RU2804029C9 - Genomic infrastructure for local and cloud processing and analysis of dna and rna - Google Patents

Abstract

FIELD: bioinformatics.

SUBSTANCE: systems, devices and methods for implementing bioinformatic protocols for performing one or more functions for analysing genome data in an integrated circuit. A method, a non-volatile computer-readable medium, a system, and a platform for analysing genome data involves one or more computers performing operations including: receiving, via an application programming interface (API), the first data identifying a particular genome data processing pipeline entered through a graphical user interface; configuring, using the Programmable Logic Device API circuits, to execute the genome data processing pipeline based on the said first data; obtaining the second data representing a genome data set or a data set derived from the genome data; one or more computers using the said genomic data processing pipeline configured in the programmable logic circuits based on the first data to process the received second data; obtaining the resulting data and providing output data based on them.

EFFECT: faster data processing with greater percentage accuracy.

22 cl, 21 dwg, 2 tbl

Description

CROSS REFERENCE TO UNRELATED APPLICATION

[0001] This application claims priority to U.S. Provisional Application No. 62/277,445, filed January 11, 2016, the contents of which are incorporated herein by reference in their entirety.

TECHNICAL FIELD

[0002] The subject matter described herein relates to bioinformatics, and more particularly to systems, devices and methods for implementing bioinformatics protocols, for example, for implementing one or more functions for analyzing genomic data on an integrated circuit, for example, a hardware processor platform .

BACKGROUND OF THE ART

[0003] Healthcare researchers and practitioners are faced with the challenge of improving the safety, quality and effectiveness of healthcare services for every patient. Personalized medicine aims to address these challenges at the individual level. For example, “genomics” and/or “bioinformatics” are fields whose goal is to ensure the safety, quality and effectiveness of preventive and therapeutic agents and treatments at a personalized, individual level. Accordingly, by applying genomics and/or bioinformatics techniques, the genetic characteristics of an individual, such as the composition of her or his genes, can be determined, and this information can be used to develop drugs and/or preventive regimens, including drug treatments personalized for that individual, allowing tailor health services to the needs of each individual.

[0004] The desire to provide personalized healthcare services is transforming the healthcare system. This transformation of the healthcare system appears to be based on disruptive innovations at the intersection of medical science and information technology, exemplified by the fields of genomics and bioinformatics. Accordingly, genomics and bioinformatics form the foundation on which this future will be built. Since the complete human genome was first sequenced in 2000 at a cost of more than $1 billion, science has advanced significantly. Today, we are on the cusp of high-resolution sequencing at a cost of less than $1,000 per genome, making it economically feasible to expand beyond the research laboratory into the broader health care landscape. Accordingly, genomic data can make critical contributions to diagnostic screening, development of preventive and/or therapeutic drugs, and/or disease treatment.

[0005] More specifically, genomics and bioinformatics are fields that apply information technology and computer science to the field of molecular biology. In particular, bioinformatics methods can be used to process and analyze a variety of genomic data, such as the genome data of an individual subject, to obtain qualitative and quantitative information about this data, which specialists in various fields can then use to develop methods of prevention and therapy to prevent or at least alleviating disease states, and thus improving the safety, quality and effectiveness of health care at the level of individual subjects.

[0006] Because bioinformatics aims to provide personalized healthcare services, it is driving the development of personalized healthcare that is proactive rather than reactive, empowering patients to take a more active role in improving their quality of life. First, federal management could be established to support research aimed at these individual aspects of disease and disease prevention; the primary challenge may be to tailor diagnostics and preventative interventions appropriate to the unique genetic profile of each individual subject. Additionally, a “network of networks” can be created to collect health-related data to help researchers determine patterns and identify genetic “patterns” of existing diseases.

[0007] The advantage of applying bioinformatics technologies in these situations is that they allow qualitative and/or quantitative analysis of molecular biological data on larger sets of samples at significantly higher speed and often greater accuracy, which contributes to the rapid development of personalized healthcare .

[0008] Accordingly, in various embodiments, molecular data for processing in bioinformatics-based platforms is typically associated with genomic data, such as deoxyribonucleic acid (DNA) and/or ribonucleic acid (RNA) data. For example, a well-known method for generating DNA and/or RNA data involves DNA/RNA sequencing. DNA/RNA sequencing may be performed manually, such as in a laboratory, or may be performed using an automated sequencer, such as at a sequencing facility, to determine the genetic makeup of a sample of a subject's genetic material, such as DNA and/or RNA. The subject's genetic information can then be used for comparison with a reference such as a reference sequence, haplotype, or theoretical haplotype to determine differences from it. Such difference information can then be used to determine or predict the occurrence of a disease state in a subject.

[0009] For example, manual or automated DNA/RNA sequencing can be used to determine the sequence of nucleotide bases in a DNA/RNA sample, such as a sample taken from a subject. Using various bioinformatics procedures, these sequences can then be combined together to obtain the subject's genomic sequence. This sequence can then be compared to a reference genomic sequence to determine if the subject's genomic sequence differs from the reference. This process involves identifying differences (variants) in a sample sequence and represents a major challenge in bioinformatics techniques.

[0010] For example, a major challenge in DNA sequencing is the construction of full-length genomic sequences, such as chromosomal sequences, from a sample of genetic material that can be compared to a reference genomic sequence to determine differences in the full-length genomic sequence from the sample. In particular, the methods used in sequencing protocols do not produce the full-length sequences of the chromosomes contained in a DNA sample.

[0011] Instead, sequence fragments are obtained that are typically 100-1000 nucleotides in length, without any indication of which genomic region they correspond to. Thus, to obtain full-length genomic chromosomal constructs, these DNA sequence fragments must be mapped, aligned, spliced, and/or compared to a reference genomic sequence. Such processes make it possible to determine differences between genomic sequences from samples and reference genomic sequences.

[0012] However, the human genome consists of approximately 3.1 billion base pairs, and each sequence fragment is typically only 100 to 500 nucleotides in length, so the time and resources required to construct and distinguish full-length genomic sequences are very large. which often necessitates the use of several different computing resources using several different algorithms over a long period of time.

[0013] In one specific example, thousands to millions of DNA sequence fragments are generated, aligned, and combined to produce a genomic sequence that approximates the length of a chromosome. One step in this process may involve comparing DNA fragments to a reference sequence to determine the region of the genome to which it aligns (i.e., to which it corresponds).

[0014] Several such steps are involved in constructing chromosome-length sequences and identifying variants (i.e., differences from the reference sequence) of a sample sequence. Accordingly, a large number of different methods have been developed to implement these stages. For example, there are widely used software that implements one or a number of stages of a bioinformatics system. However, a common feature of such bioinformatics methods and software-based systems is that they are labor intensive, take a long time to run on conventional processors, and are prone to errors.

[0015] Accordingly, a bioinformatics system that could implement algorithms implemented in such software in a less labor-intensive and/or less processing-intensive manner with greater percentage accuracy would be useful. However, even as we approach the cost of $1,000 Genome analysis, storing and sharing this raw digital data far outstrips the cost of obtaining it. This analysis bottleneck is a major barrier to moving from this ever-growing raw data to the real medical knowledge we seek to gain from it.

[0016] Accordingly, systems, devices and methods are provided herein for applying genomics and/or bioinformatics protocols, for example, to implement one or more functions for analyzing genomic data, for example, using software implementations and/or on an integrated circuit, for example ,on the processing hardware platform. For example, as described below herein, in various embodiments, combinations of software and/or hardware acceleration solutions, such as solutions including integrated circuits and software to interface with them, can be used to solve bioinformatics-related problems when In this case, integrated circuits may consist of one or more hardwired digital logic circuits that may be interconnected by a plurality of physical electrical connections, which may be organized as a set of processing modules, each processing module may be configured to perform one or more steps bioinformatics protocol for genetic analysis. The advantage of such an organization is that these bioinformatics tasks can be completed more quickly than using software alone, as is usually the case for solving such problems. However, the application of such hardware acceleration technique is not currently commonly applied in the fields of genomics and/or bioinformatics.

SHORT DESCRIPTION

[0017] The present invention relates to performing tasks such as tasks that are part of bioinformatics protocols. In various embodiments, multiple tasks are performed, and in some cases, these tasks are performed in such a way that they form a pipeline, with each task and/or its near completion acting as a building block for each subsequent task until the desired end result is achieved. Accordingly, in various embodiments, the present invention relates to implementing one or more methods on one or more devices, where the device is optimized to implement the methods. In some embodiments, the one or more methods and/or one or more devices are packaged into one or more systems.

[0018] For example, in some aspects, the present invention relates to systems, devices and methods for implementing genomics and/or bioinformatics protocols, such as, in various embodiments, for implementing one or more functions for obtaining and/or analyzing genetic data using innovative software and/or on an integrated circuit, for example, implemented in a combination of software and/or hardware processing platform. For example, in one aspect, a genomic and/or bioinformatics system is provided. This system may include performing various functions to obtain bioanalytical data and/or analysis that are optimized for faster implementation and/or increased accuracy. Methods for implementing these functions can be implemented in software or hardware solutions. Accordingly, in some embodiments, methods are provided that include obtaining and/or collecting data and/or analysis, which may include implementing one or more algorithms, wherein said algorithm(s) are optimized in accordance with, for example, software, hardware, or the combination of hardware and software in which they are implemented. In particular, in cases where the algorithm is implemented in a software solution, the algorithm and/or its constituent processes may be optimized to run faster and/or run with increased accuracy when implemented using a particular tool. Likewise, in cases where an algorithm is implemented in a hardware solution, the hardware is designed specifically to implement those functions and/or their constituent processes, optimized for faster execution and/or execution with increased accuracy when implemented using a particular tool. Further, where functions involve a combination of software and/or hardware solutions, those functions and their constituent processes are designed and configured to work together optimally, achieving previously unattainable speeds with the same or increased accuracy.

[0019] Accordingly, in one aspect of the present invention, systems, devices and methods are provided for implementing bioinformatics protocols, for example, for performing one or more functions for generating and/or analyzing genetic data, for example, using one or more developed and/or optimized algorithms and/or on one or more optimized integrated circuits, for example, on one or more hardware processing platforms. Accordingly, in one embodiment, methods are provided for implementing one or more algorithms for performing one or more steps for generating and/or analyzing genomic data in a genomics and/or bioinformatics protocol. In another case, methods are provided for implementing the functions of one or more algorithms for performing one or more steps for analyzing genomic data in a bioinformatics protocol, wherein the functions are at least partially implemented on an integrated circuit, such as a circuit consisting of one or more hardware-connected digital logic gates. schemes In such an embodiment, hardwired digital logic circuits may be coupled, for example, through a plurality of physical electrical connections, and may be arranged to operate as one or more processing modules. In various embodiments, a plurality of hard-wired digital logic circuits are provided, the hard-wired digital logic circuits being configured as a set of processing modules, each processing module being capable of executing one or more stages of bioinformatics genetic analysis protocols, as in the case of a bioinformatics processing pipeline.

[0020] More specifically, in one embodiment, a system is provided for obtaining genetic sequence data, for example, including devices and methods for sequencing nucleic acids, and/or for implementing a sequence analysis pipeline on such genetic sequencing data. The system may include one or more electronic data sources, such as those associated with a DNA/RNA sequencing device, such as those described herein, a memory and/or an integrated circuit. For example, one embodiment includes an electronic data source, wherein said electronic data source may be configured to generate and/or provide one or more digital signals, such as digital signals representing one or more genetic data reads, wherein, for example, each genomic data read data includes a nucleotide sequence. Further, the memory may be configured to store one or more reference genetic sequences, and may further be configured to store an index, such as an index of the one or more reference genetic sequences, and/or annotated splice boundary data.

[0021] Further, a device and/or method for obtaining genetic sequence data is provided. For example, an approach to DNA/RNA analysis, for example, for genetic diagnosis and/or sequencing, is provided, comprising one or more reactions of hybridization, detection, and/or nucleic acid sequencing. In various embodiments, the approach may include hybridization and/or detection devices and/or procedures for performing one of the following steps. In particular, for purposes of genetic analysis, a sample of the subject's TDT RNA DNA can be isolated and immobilized, for example, directly and/or indirectly, on a substrate, such as a substrate containing a chemically sensitive one-dimensional (1-D) and/or two-dimensional (2D) reaction layer , for example, a graphene reaction layer, and/or a three-dimensional (3D) reaction layer, and a probe of a known or detectable genetic sequence, such as a disease marker, can be spilled onto the surface, or vice versa. In various embodiments, one or more of the RNA or DNA samples and/or probe may contain a label.

[0022] In other embodiments, such as when the substrate includes a 1D or 2D, for example, graphene, reaction layer, and/or other chemically sensitive reaction layer, label or probe, such a chemical or radioactive label is not necessarily necessary and/ or present. In either embodiment, if a disease marker is present, a binding event, such as hybridization, will occur, and since binding events can be detected, for example, by labeling the analyte or probe and/or through appropriate reaction layer organization as described herein, the presence of the marker diseases can also be detected. In the case where the disease marker is not present, there will be no reaction and, accordingly, the marker will not be detected. Of course, in some cases, the indicative event may be the absence of a binding event. Accordingly, the system may be configured such that a hybridization event may or may not be detected, indicating the presence or absence of a disease marker in a sample taken from a subject.

[0023] Similarly, for DNA and/or RNA sequencing, first an unknown nucleic acid sequence to be identified, such as a single-stranded DNA or RNA sequence of a subject, is isolated, amplified, and immobilized on a support, which, as described herein, may include 1D-, 2D-, for example, graphene, layer, 3D, or other reaction layer deposited on it. A known nucleic acid, for example a nucleotide base that can be labeled with an identifiable label, is then brought into contact with an unknown nucleic acid sequence in the presence of a polymerase. As noted above, in the event that a reaction event occurs in the vicinity of a suitably configured reaction layer, for example a graphene-containing reaction layer, the labeled reagent may not be used.

[0024] Accordingly, when hybridization occurs, the nucleic acid binds to the appropriate complementary base in an unknown sequence, for example, in DNA or RNA from a sample being sequenced, and is immobilized on the surface of the support, for example, near the reaction layer. The binding event can be detected, for example, by optical means, electrical means, and/or through a suitably detectable reaction occurring on the reaction layer. These steps are then repeated until the DNA or RNA from the sample has been completely sequenced. Typically, these steps are performed on a next generation sequencer in a manner known in the art, or they can be performed in accordance with the devices and methods described herein, allowing thousands to millions of sequencing reactions to be conducted and/or processed simultaneously, and the digital data obtained the resulting reactions can be analyzed using the innovative sequencing devices and methods disclosed herein, for example, in a multiplex bioinformatics processing pipeline.

[0025] For example, in one aspect, such as those associated with innovative sequencing devices presented herein, a suitably configured platform can be implemented as a field-effect transistor (FET) containing a chemical reaction layer, such as a transistor for use in a reaction hybridization and/or sequencing. In particular, such a field effect transistor (FET) can be implemented on a primary structure such as a substrate, for example a silicon substrate. In various embodiments, the primary structure may include one or more additional structures, for example, in a stacked configuration, for example, in the form of a layer of insulating material. For example, the insulator material may be located over the primary structure of the silicon substrate, and may be an inorganic material such as silica or silica, or an organic material such as a polyimide, benzocyclobutene or the like.

[0026] The primary structure and/or insulator layer may include an additional structure containing one or more charge carrier sources and/or charge carrier sinks, for example, separated from each other by some space, and integrated into the primary structure and/or layer of insulating material , and/or may lie in the same plane with the upper and/or lower surface of the insulator to form an upper and/or lower gate. In various embodiments, the structures, for example, a silicon substrate structure, may further include, or may be associated with, an integrated circuit, such as a processor, e.g., a microprocessor, for processing generated data, such as sensor data, e.g., reaction data sequencing, for example, near the gate area. Accordingly, the plurality of structures may be in the form of, or include, an integrated circuit, and/or may be in the form of an ASIC (special purpose integrated circuit), structured ASIC, or FPGA (field programmable gate array).

[0027] In particular, these structures may be in the form of a complementary metal-oxide-semiconductor (CMOS) structure, which in turn may be in the form of a chemically sensitive field-effect transistor sensor containing one or more charge carrier sources, a charge carrier sink and/or a reaction region such as a gate, which may itself include a micro- or nanochannel, chamber and/or well configuration, which sensor may be adapted to interface with a processor. For example, a field-effect transistor may include a CMOS configuration comprising or associated with an integrated circuit that is configured on a silicon substrate that further includes an insulator layer, wherein said insulator layer includes a charge carrier source and a charge carrier drain integrated into the insulator layer, the source and drains can be made of metal such as inlaid copper. In various embodiments, CMOS and related structures may include a surface, such as a top surface, that may include a channel and/or a chamber to form a reaction well, wherein the surface of the reaction well may extend beyond a charge carrier source to a charge carrier sink and may be adapted for receiving various reagents used in biochemical reactions, such as hybridization and/or DNA or RNA sequencing reactions.

[0028] In some embodiments, the surface and/or channel and/or chamber may include a one-dimensional transistor material, a two-dimensional transistor material, a three-dimensional transistor material, and/or the like. Various embodiments may include a one-dimensional (1D) transistor material, where the 1D material may consist of a carbon nanotube or semiconductor nanowire, which in various embodiments may be in the form of a sheet or channel, and/or in various embodiments may include a nanopore, although in many embodiments In variants, the nanopore is absent or not necessary. Some embodiments may comprise a two-dimensional (2D) transistor material, where the 2D material may include a graphene layer, silicene, molybdenum disulfide, black phosphorus and/or configuration. In various embodiments, the surface and/or channel may include a dielectric layer. Additionally, in various embodiments, a reaction layer, e.g., an oxide layer, may be located on the surface and/or in a channel and/or chamber, e.g., layered or otherwise placed on a 1D, 2D, e.g., graphene 3D layer ( layers). Such an oxide layer may be an aluminum oxide or a silicon oxide, for example silica. In various embodiments, the passivation layer may be located on the surface and/or channel and/or chamber, for example, it may be layered or otherwise deposited on a 1D, 2D, e.g., graphene, or 3D layer(s) ) and/or on the associated reaction layer on the surface and/or channel and/or chamber.

[0029] In particular embodiments, the primary and/or secondary and/or tertiary structures may be fabricated or otherwise configured to include a well or chamber structure in and/or on the surface, for example, to form a reaction region. For example, the well or chamber structure may be located on a portion of the surface, such as the outer surface of the primary and/or secondary and/or tertiary structures. In some embodiments, the well-containing structure may be in the form of a micro- or nanochamber and may be provided on top, or may include at least a portion of a one-dimensional, two-dimensional, e.g., graphene, and/or three-dimensional material, and/or further include a reaction material, e.g. , oxide and/or passivation layers. In various embodiments, the structure containing the chamber and/or well may define an opening, such as an opening that provides access to the interior of the chamber, such as allowing direct contact with a one-dimensional, such as a carbon nanotube or nanowire, a two-dimensional, such as a graphene, or a three-dimensional surface and/or channel and/or chamber. In certain embodiments, the chamber and/or well may be sized to be defined as a micro- or nano-chamber.

[0030] Accordingly, one additional aspect of the present invention is a biosensor, for example, a biosensor for performing a nucleic acid sequencing reaction. The biosensor includes a CMOS structure, which may be a chemically sensitive FET sensor, and may include a metal-containing source and drain, such as a source and/or drain, which further includes a sequence, such as a reaction zone, which includes a sequence with a one-dimensional or two-dimensional layer, such as a graphene layer, or a 3D sequence that extends from source to drain. In particular, the reaction region may comprise or be configured as a well or chamber structure that may be located on a portion of the outer surface of the well with a 1D or 2D layer. In such an embodiment, the well structure may be designed to define an opening that allows direct contact with the nanotube, nanowire, and/or graphene surface of the well or chamber. In various embodiments, the oxide and/or passivation layer may be located in or on the surfaces of the chamber. Accordingly, in some embodiments, a chemically sensitive transistor, such as a field-effect transistor (FET), may be provided including one or more nano- or microwells for sequencing reactions.

[0031] In some embodiments, the chemically sensitive field effect transistor may include multiple wells and may be configured as an array, such as a sensor array. Such matrix or matrices may be configured to detect the presence and/or changes in concentration of various types of analytes in a wide range of chemical and/or biological processes, including DNA and/or RNA hybridization reactions and/or DNA or RNA sequencing. For example, the devices and/or systems incorporating such devices described herein can be used in a method for analyzing biological or chemical materials, for example, whole genome sequencing and/or analysis, genotyping, microarray analysis, panel analysis, exome analysis, microbiome analysis and/or clinical analysis, such as cancer analysis, NIPT analysis (non-invasive perinatal diagnosis) and/or uterine carcinosarcoma analysis, etc.

[0032] Accordingly, in one particular embodiment, a graphene field-effect transistor (gFET) array can be used to implement DNA and/or RNA sequencing techniques and process this data, for example, in the genetic analysis pipeline described herein. For example, a field-effect transistor array technology such as graphene field-effect transistor (gFET) technology can be configured to include reaction wells with a reaction layer, with the ability to detect changes in the concentration of hydrogen ions (pH), changes in the concentrations of other analytes, and/or binding events associated with chemical processes, such as processes associated with DNA or RNA synthesis, for example, in a controlled reaction chamber or gFET-based sensor well. Such a chemically sensitive field effect transistor may include, or may be adapted to be coupled to, one or more integrated circuits and/or adapted to enhance the measurement sensitivity and/or accuracy of the sensor and/or associated array(s), for example, by including one or more sequences in a reaction chamber or reaction well, wherein at least one sequence is coated with a layer of one-dimensional (1D) and/or two-dimensional (2D) and/or three-dimensional (3D) material, a dielectric reaction layer, a passivation layer, and/or the like.

[0033] Accordingly, one aspect of the present invention may include one or more integrated circuits, which may consist of one or more sets of hardware-connected digital logic circuits, as in the case where a set of hardware-connected digital logic circuits has interconnections, such as, for example, through a variety of physical electrical interconnections, and may be adapted to perform and/or detect hybridization and/or DNA or RNA sequencing reactions, such as primary processing, and/or may be further adapted to implement the results thereof, such as in one or more stages of secondary and/or tertiary processing. In such embodiments, the integrated circuit may include an input, such as one or more of a plurality of physical electrical interconnects, that allows connection to an electronic data generation source, such as a CMOS FET sequencing device according to the present invention and/or a Next Gen Sequencer ) configured to generate such data, for example, in the form of multiple sequenced segments, eg reads, genomic data. In specific embodiments, the one or more integrated circuits may include a set of hardware-connected digital logic circuits that are configured to implement a secondary and/or tertiary processing pipeline for the generated genomic data reads, for example, through one or more interconnects.

[0034] In this case, fixed digital logic circuits and/or interconnects can be configured to obtain one or more genomic data reads, for example, from an electronic data source. In specific embodiments, one or more hardwired digital logic circuits may be organized as a set of processing modules, for example, a set of processing modules, where each processing module is composed of a subset of hardwired digital logic circuits and is configured to execute one or more stages of a sequencing and/or analysis pipeline, for example, multiple reads of genomic data. In such embodiments, each subset of hardwired digital logic circuits may, in some embodiments, have a hardwired configuration enabling one or more steps in the sequencing and/or analysis pipeline. However, as noted above, one or more steps in the sequencing and/or analysis pipeline may be implemented in software, such as when the software and/or hardware are adapted to interact with each other in an optimized manner during operation.

[0035] Accordingly, in various embodiments, a plurality of hardwired digital logic circuits are provided, wherein said hardwired digital logic circuits are organized in the form of a set of processing modules, wherein one or more processing modules includes one or more sequencing modules and/or modules mapping (mappers) and/or alignment modules and/or sorting modules (sorters) and/or variant determination modules and/or tertiary processing modules described in this document. For example, in various embodiments, the one or more processing modules may include a mapping module, wherein the mapper may be electrically coupled and further configured to interface with a memory on the device or otherwise coupled thereto, e.g. through a suitably configured interconnect providing access to an index containing one or more reference genetic sequences, one or more reads generated from sequencing, and/or an index of splice boundaries (for example, in the case of RNA sequencing), wherein the mapper is used to performing one or more mapping operations.

[0036] In particular, a suitably configured processing module or modules may include, or may be adapted to be used as a mapping module, to perform one or more mapping operations, for example, including accessing an index of one or more reference genetic sequences from memory, for example, through one or more of a plurality of physical electronic interconnections, for example, so as to map a plurality of reads to one or more to one or more segments of one or more reference genetic sequences. Additionally, in various embodiments, the one or more processing modules may include an alignment module that may be wired to memory and may be configured to evaluate one or more reference genetic sequences from memory, for example, through one or more of a plurality of physical electronic interconnects, for example, with the ability to align multiple reads to one or more segments of one or more reference genetic sequences.

[0037] Further, in various embodiments, said one or more processing modules may include a sort module, wherein said sort module may be wired and configured to evaluate one or more aligned reads from memory, for example, by one or more of a plurality of physical electronic interconnects, for example, with the ability to sort each aligned read, for example, according to one or more positions of that read in one or more reference genomic sequences. In such embodiments, said one or more of the plurality of physical electrical interconnections may include an integrated circuit output, for example, for communicating resultant data from a mapping module and/or an alignment module and/or a sorting module. Further, in specific embodiments, as discussed above, one or more processing modules may be configured to interact with various software-implemented functions, for example, through one or more interconnections, for example, a plurality of physical electronic interconnections, to implement one or more steps in the pipeline analysis, including implementation of one or more RNA and/or DNA sequencing protocols and/or variant detection protocol.

[0038] In various embodiments, the plurality of one or more electronic circuits may include a main controller to provide a wired configuration for each subset of hardware-connected digital logic circuits, for example, to perform one or more mapping, alignment, and/or sorting functions, wherein said functions may be implemented as one or more stages of a sequence analysis pipeline and/or may include performing one or more aspects of sequencing and/or variant detection functions. Further, in various embodiments, the one or more integrated circuits disclosed herein may be implemented as a field programmable gate array (FPGA) containing hardware-interconnected digital logic circuits, as in the case where the interconnected configuration can be obtained after manufacture integrated circuit, and, accordingly, can be non-volatile. In other various embodiments, an application-specific integrated circuit (ASIC) comprising hardware-interconnected digital logic circuits is implemented. In various other embodiments, the integrated circuit may be configured as a special purpose structured integrated circuit (Structured ASIC) containing hardwired digital logic circuits.

[0039] In some embodiments, the one or more integrated circuits, such as CMOS FET sequencing chips, and/or a biosensor, and/or one or more attachable storage devices, may be located on an expansion board, such as a PCI extensions, for example, in various embodiments, the integrated circuit(s) of the present invention may be a chip(s) with a PCIe card. In various embodiments, the integrated circuit and/or chip may be components of a sequencer, such as an automated sequencer that uses FET sensors and/or a next generation sequencer (NGS), and/or, in other embodiments, may be capable of The integrated circuit and/or expansion board can be accessed via the Internet, for example, via the cloud. Further, in some embodiments, the storage device may be volatile RAM (RAM) or dynamic RAM (DRAM).

[0040] Accordingly, in one aspect, an apparatus is provided for performing one or more steps of a sequence analysis pipeline, such as genetic data analysis, wherein the genetic data includes one or more of a reference genetic sequence(s), an index of one or more genetic sequence(s) , an index of one or more reference genetic sequence(s), an index of one or more splice boundaries, such as an annotated index or splice boundary table, and/or a set of reads, such as genetic data, such as DNA or RNA. In various embodiments, the device may comprise an integrated circuit, wherein said integrated circuit may include one or more, for example, a set of hardwired digital logic circuits, wherein said set of hardwired digital logic circuits can be connected by interconnections, for example, a plurality of physical electrical interconnections. In some embodiments, said one or more of a plurality of electrical interconnections may include an input, such as an input to obtain multiple reads of genomic data, for example, from a sequencing device, as described herein. Additionally, a set of hardware-connected digital logic circuits may be further implemented in a connected configuration with access to an index of one or more reference genetic sequences and/or annotated splice boundaries, through one of a plurality of physical electrical interconnects, and with the ability to map multiple DNA reads and/or RNA into one or more segments of one or more reference genetic sequences, for example, according to an index or indexes.

[0041] In various embodiments, the index may include one or more hash tables, such as a primary and/or secondary and/or splice boundary table. For example, a primary hash table may be present, in which case a set of hardwired digital logic circuits may be configured to perform one of the following operations: extract one or more genetic data seeds from a plurality of genetic data reads; performing a primary hashing function, for example, with one or more genetic data seeds, generating a reference address for each of the one or more seeds, and evaluating a primary hash table using the reference address to obtain a position in one or more reference genomic sequences for each from one or more genetic data seeds. In various embodiments, said one or more genetic data seeds may contain a fixed number of nucleotides.

[0042] Further, in various embodiments, the index may include a secondary hash table, for example, as in the case where a set of hardwired digital logic circuits is configured to at least one of the following: extend at least one of one or more seeds for counting additional adjacent nucleotides to produce at least one extended genetic data seed; performing a hash function, such as a secondary hash function, on the at least one extended genetic data seed, generating a second reference address for the at least one extended seed, and evaluating a secondary hash table, such as using the second reference address, obtaining a position in one or more reference genetic sequences for each of the one or more extended genetic data seeds. In various embodiments, the secondary hashing function may be performed by a set of hardwired digital logic circuits, such as when the primary hash table announces an extension entry, which is an instruction to the hardwired digital logic circuitry to extend one or more seeds with additional adjacent nucleotides. In some embodiments, the extension entry may indicate the number of additional adjacent nucleotides with which the at least one primer should be extended, and/or the manner in which the primer should be extended, for example, by an equal number of “x” nucleotides at each end of the primer in the same manner.

[0043] Further, DNA is known to encode genes. However, gene expression requires translation and transcription of the genetic code into proteins. In particular, a gene can be transcribed in the cell nucleus by RNA polymerase enzymes into a transcript - messenger RNA (mRNA) or other types of RNA (for example, transfer RNA). The RNA transcript itself is a single-stranded copy of the gene, except that the base thymine (T) present in the DNA is replaced by uracil (U). However, immediately after the formation of this copy, its sequence includes copies of both various exons and introns, and the various copies of introns are usually excised during the splicing process, for example, by spliceosomes, leaving only copies of exons that must be connected to each other at “junctions” (points, boundaries) of splicing” (which after this cannot always be easily identified), with the formation of codon regions. The spliced mRNA containing the codon regions is then transported from the cell nucleus to the ribosome, which transcribes it into protein, with each group of three RNA nucleotides forming a codon that codes for one amino acid. During the decoding process, amino acids are linked into chains, and as a result of binding and glycosylation, proteins are formed that make up the cells, tissues and organs of the body. In this way, genes in DNA function as the initial instructions for producing proteins.

[0044] Accordingly, since DNA includes both coding regions, such as exons, and non-coding regions, such as introns, mapping and/or alignment and/or sorting of RNA relative to the original genetic precursor in genomic DNA can be difficult. Specifically, each gene exists on a single strand of a double-stranded DNA double helix, often as a series of exons (coding segments) separated by introns (non-coding segments). Some genes contain a single exon, but most contain several exons (separated by introns), and some contain hundreds of exons or thousands of exons. Exons are usually several hundred nucleotides in length, but they can be shorter, up to one nucleotide, or longer, up to tens, hundreds, or thousands of nucleotides. The length of introns is usually thousands of nucleotides, and in some it exceeds a million nucleotides. Accordingly, when mapping, aligning, and/or sorting into RNA, such as spliced mRNA, portions of that spliced mRNA may come from different regions of DNA that may be separated from each other by one, or two, or even a million or more nucleotides. This makes RNA processing highly complex.

[0045] However, one aspect of the present invention overcomes these difficulties through the methods disclosed herein and accordingly allows for rapid and accurate whole transcriptome sequencing, mapping, alignment and/or sorting of RNA. More specifically, in embodiments involving RNA processing, the above index may include one or more tables, such as a hash table or other index, including or associated with a table that allows easy access to various known or defined splice boundaries used in biological systems when transcribing RNA from DNA, as described in detail below. Accordingly, in such embodiments, the RNA-competent mapping/sorting module may be configured to process such splice boundaries and match RNA sequence reads to segments of transcribed and spliced RNA, such as when the read crosses one or more splice boundaries; which, in relation to a DNA-based reference genome, means that the first part of the read comes from the first exon and should map to the first exon, and the second part of the read comes from the second exon and should map to it, and so on. Accordingly, the index may include or be associated with one or more splice boundary tables, and a set of hardwired digital logic circuits may be configured to perform one of the following operations: apply said splice boundary data to determine and/or retrieve one or more genetic seeds data, such as RNA, from multiple RNA-based genetic data reads; performing a function, for example, a hashing function, for example, on said one or more RNA genetic data seeds, generating a reference address for each of said one or more seeds; and evaluating a hash table using the reference address to obtain a position in one or more reference genomic sequences for each of the one or more RNA genetic data seeds.

[0046] Additionally, in one aspect, an apparatus is provided for performing one or more sequence analysis pipeline steps on genetic sequence data, such as DNA or RNA, wherein said genetic sequence data includes a single genetic sequence or a plurality of genetic sequences, which may include both exons, and introns, an index of one or more reference genetic sequences and/or an index of annotated splice boundaries, and a set of genomic reads. In various embodiments, the device may include an integrated circuit, wherein said integrated circuit may include one or more, for example, a set of hardwired digital logic circuits, wherein said set of hardwired digital logic circuits can be connected by interconnections, for example, a plurality of physical electrical interconnections. In some embodiments, said one or more of a plurality of physical electrical interconnections may include an input, such as an input to obtain a plurality of genomic data reads, which reads may be preprocessed as described herein, eg, mapped. Additionally, a set of hardwired digital logic circuits may be further coupled to provide access to one or more reference genetic sequences, through one of a plurality of physical electrical interconnections, to obtain position information, for example, from a mapper, indicating one or more segments of one or more reference sequences, and for aligning multiple reads to said one or more segments of one or more reference genetic sequences.

[0047] Accordingly, in various embodiments, a set of wired digital logic circuits is configured to align a plurality of DNA RNA genetic data reads to one or more segments of one or more reference genetic sequences, and further includes a waveform processing module, which may be formed by a connected configuration of a set of hardware-connected digital logic circuits. In some embodiments, said waveform processing module may be configured to process an array of alignment matrix elements, such as a matrix defined by a subset of a set of hardwired digital logic circuits. For example, in some embodiments, the alignment matrix may define a first axis, for example, representing one of a plurality of reads, and a second axis, for example, representing one or more segments of one or more reference genetic sequences. In such an embodiment, said waveform processing module may be configured to generate a wave front profile of dimples that are located along an element array from a first axis to a second axis, and may also be configured to generate a score, for example, a score for each element in the element waveform profile, wherein said score may represent the degree of match between said one of the plurality of reads and one or more segments of one or more reference genetic sequences.

[0048] In such an embodiment, said waveform processing module may be further configured to move the wavefront profile of the elements across the alignment matrix such that the highest score may be centered on the wavefront profile of the wells. Additionally, in various embodiments, said wave processing module may be further configured to track back one or more, for example, all positions in the wave profile of elements assigned scores by preceding positions in the alignment matrix; tracking one or more, for example, all tracked trajectories to convergence, and generating a CIGAR string based on the convergence trace.

[0049] In some embodiments, a connected configuration of a set of hardwired digital logic circuits for aligning multiple reads to one or more segments of one or more reference genetic sequences may include a connected configuration for applying the Burrows-Wheeler algorithm as described above, for example, for mapping before alignment, and/or to implement the Smith-Waterman and/or Needleman-Wunsch estimation algorithm. In such an embodiment, the Smith-Waterman and/or Needleman-Wunsch scoring algorithm may be configured to implement a scoring parameter that is sensitive to the quality scores of the grounds. Further, in some embodiments, the Needleman-Wunsch estimation algorithm may be an affine Needleman-Wunsch algorithm.

[0050] In specific embodiments, a device may include an integrated circuit, wherein said integrated circuit may include one or more, for example, a set of hard-wired digital logic circuits, wherein said set of hard-wired digital logic circuits can be interconnected by, for example, a plurality of physical electrical interconnections. In some of these embodiments, said one or more of a plurality of physical electrical interconnections may include an input, such as an input to obtain a plurality of reads of genomic data, wherein the reads may be preprocessed as described herein, for example, mapped and/or aligned. Additionally, a set of hardwired digital logic circuits may further be in an interconnected configuration that provides access to one or more genetic sequences, through one of a plurality of physical electrical interconnections, to obtain position information, for example, from a mapping module and/or an alignment module , which points to one or more segments of one or more reference sequences, and for sorting multiple reads with one or more segments of one or more reference genetic sequences.

[0051] Accordingly, in one aspect, a method may be provided for sequencing genetic material, for example, to obtain electronic genetic data. In specific embodiments, the method includes using a next generation sequencer to sequence genomic DNA and/or corresponding RNA, as described generally herein and as known in the art. In other embodiments, the method includes using a next generation sequencer modified as described herein to sequence genomic DNA and/or corresponding RNA. In other embodiments, the method involves using a FET and/or CMOS sequencer, such as an on-chip sequencer as described in detail herein, to sequence genomic DNA and/or corresponding RNA. In various embodiments, the genetic material, once received, can be converted into electronic form, such as digital form, which can be streamed or otherwise transferred to one or more pipeline modules described herein.

[0052] Additionally, after receiving electronic, eg analog or digital genetic data, such as sequencing data, another aspect of the present invention relates to performing sequential sequence analysis pipeline operations on such sequencing data of genetic material. This genetic data may include one or more reference genetic sequences, one or more iodine index or more reference genetic sequences, and/or a list of one or more annotated splice boundaries (e.g., in the case of RNA sequencing) associated therewith, and/or a set of reads genomic data (e.g. DNA and/or RNA). The method may include one or more different iterations of receiving, evaluating, mapping, aligning, and/or sorting said genetic sequence data. For example, in some embodiments, a method may include receiving, as input to an integrated circuit from an electronic data source, one or more reads of genomic data, wherein each genomic data may include a nucleotide sequence. In such an embodiment, the integrated circuit may be formed by a set of hardware-connected digital logic circuits, for example, connected by a plurality of physical electrical interconnections, which electrical interconnections may include one or more of a plurality of physical electrical interconnections including an input.

[0053] The method may further include estimating, by the integrated circuit at one or more of a plurality of physical electrical interconnects from the storage device, an index of one or more reference genetic sequences and/or, in the case of RNA sequencing, annotated splice boundaries. In particular, if annotated splice boundaries are supplied to the mapping module, they can be used to improve the sensitivity of mapping. In such an embodiment, the list of annotated boundaries can be loaded into the storage device so that the mapping module can access it, thereby facilitating the mapping of RNA genetic material. Preferably, the annotated boundaries may be in the form of a table, such as a hash table or index, which may be associated with it so that it can be easily accessed by the mapping module. Accordingly, the method may include mapping, by a first subset of hardware-connected digital integrated circuit logic circuits, a plurality of genetic reads, such as DNA or RNA reads, one or more segments at a time of one or more reference genetic sequences. Additionally, the method may include evaluating, on an integrated circuit, one or more of a plurality of physical electrical interconnections from a storage device, said one or more mapped reads, reads, and/or reference genetic sequences; and aligning, by a second subset of hardware-connected integrated circuit digital logic circuits, a plurality of reads, eg, mapped reads, with one or more segments of one or more reference genetic sequences.

[0054] In various embodiments, the method may further include estimating, by the integrated circuit on one or more of the plurality of physical electrical interconnects from the memory, an aligned set of reads of reads. In such an embodiment, the method may include sorting, by a third subset of hardware-connected integrated circuit digital logic circuits, an aligned set of reads according to their positions in one or more reference genomic sequences. In some embodiments, the method may further include outputting, for example, on one or more of a plurality of physical electrical interconnects of the integrated circuit, the resulting mapping and/or alignment and/or sorting data, as in the case where the resulting data includes mapped positions and/or an aligned and/or sorted set of reads.

[0055] Next, after generating and/or processing the genetic data, for example, in one or more secondary processing protocols, for example, by mapping, alignment and/or sorting, for example, producing one or more variant definition files, for example, for determining how specified genetic sequence data of a subject differs from one or more reference sequences, a further aspect of the present invention may involve performing one or more analytical functions on the generated and/or processed genetic data, for example, for further tertiary processing. For example, the system may be configured to further process generated and/or secondary processed data, for example, by processing in one or more tertiary processing pipelines, for example, one or more of a genomic pipeline, an epigenomic pipeline, a metagenomic pipeline, a co-genotyping pipeline , MuTect2 pipeline or other tertiary processing pipeline. In particular, in various embodiments, an additional level of processing may be provided, for example, for disease diagnosis, therapeutic treatment and/or prophylactic treatment, such as using NIPT, NICU, Cancer, LDT, AgBio (agricultural and biological analysis) procedures and others similar procedures for the diagnosis, prevention and/or treatment of diseases, using data generated by one or more of the presented primary and/or secondary and/or tertiary pipelines. Accordingly, the devices and methods disclosed herein can be used to generate genetic sequence data, which data can then be used to generate one or more variant definition files and/or other associated data, which can then be submitted for processing in pipelines tertiary processing in accordance with the devices and methods disclosed herein, for example, for specific and/or general diagnosis of diseases, as well as for preventive and/or therapeutic treatment and/or, for developmental applications.

[0056] Accordingly, in various embodiments, embodiments of various aspects of the present invention may include, but are not limited to: devices, systems and methods incorporating one or more of the features described in detail herein, as well as articles that include a tangible computer-readable medium configured to cause one or more machines (eg, computers, etc.) to perform the operations described herein. Likewise, computer systems and/or networks are also described that may include one or more processors and/or one or more storage devices coupled to said one or more processors directly or remotely. Accordingly, computer-implemented methods in accordance with one or more embodiments of the present invention may be implemented by one or more data processors located on a single computing system or multiple computing systems, such as one or more computer clusters. Such multiple computing systems may be coupled to and may exchange data and/or commands or other instructions and the like. through one or more connections, including, but not limited to: a connection through a network (e.g., the Internet, wireless wide area network, local area network, wide area network, wired network, etc.), or through a direct connection between one or more computing systems, etc. The memory, which may include a computer-readable medium for storing information, may contain, encode, store, and the like. one or more programs that cause the processor to perform one or more of the operations described herein.

[0057] Options for implementing the object described in this document are disclosed in detail below in the accompanying graphics and in the description below. Other features and advantages of the subject matter described herein will become apparent from the description and drawings, as well as from the claims. While certain features of the subject matter described herein are disclosed for illustrative purposes in relation to an enterprise software system or other business solution or architecture, such features are not intended to be limiting. It is assumed that the features that follow this disclosure determine the scope of the protected object.

DESCRIPTION OF GRAPHIC MATERIALS

[0058] The accompanying drawings, which are included in and constitute a part of this specification, illustrate certain aspects of the subject matter disclosed herein and, together with the description, serve to explain certain principles associated with the disclosed embodiments. In these graphic materials:

[0059] In FIG. 1 depicts an RNA read illustrating the intersection between one or more boundaries (dots) and primers covering the splice boundaries of the read.

[0060] In FIG. Figure 2 shows an example RNA read illustrating that short (L bases) primers can be configured to better fit short exons and account for exon overhangs or exon segments excised during editing, such as single nucleotide polymorphisms (SNPs).

[0061] In FIG. 3 depicts examples of reference intervals within the search range of successfully mapped K-base seeds that can be used to query a fixed seed hash table, for example, using L-base seeds.

[0062] In FIG. Figure 4 shows comparisons of parts of the reads to the left and right of the stitching position.

[0063] In FIG. 5 shows an abstract alignment rectangle with the concatenated query sequence on the vertical axis and the concatenated reference sequence on the horizontal axis.

[0064] In FIG. 6 shows a device in accordance with one embodiment of the present invention.

[0065] FIG. 7 shows another device in accordance with an alternative embodiment of the present invention.

[0066] In FIG. 8 shows a block diagram of the genomic infrastructure for local and/or cloud processing and analysis using genomics methods.

[0067] In FIG. 9 is a block diagram of local and/or cloud computing according to FIG. 8 for genomic infrastructure for local and/or cloud processing and analysis using genomics methods.

[0068] FIG. 10 is a block diagram according to FIG. 9, which shows in more detail the implementation of computing for genomic infrastructure for local and/or cloud processing and analysis using genomics methods.

[0069] FIG. 11 is a block diagram according to FIG. 8, which shows in more detail the 3rd party analytics for genomic infrastructure for local and/or cloud genomics processing and analysis.

[0070] In FIG. 12 is a block diagram illustrating a hybrid cloud computing configuration.

[0071] FIG. 13 is a block diagram according to FIG. 12 illustrating the hybrid cloud computing configuration in more detail.

[0072] FIG. 14 is a block diagram according to FIG. 13 illustrating the hybrid cloud computing configuration in more detail.

[0073] FIG. 15 is a block diagram illustrating a primary, secondary and/or tertiary analysis pipeline according to the present disclosure.

[0074] In FIG. 16 depicts a block diagram for the analysis pipeline described herein.

[0075] FIG. Figure 17 shows an example of integrated circuit design and assembly.

[0076] FIG. 18 is a block diagram of a processor hardware architecture in accordance with another embodiment of the disclosed subject matter.

[0077] FIG. 19 is a block diagram of a processor hardware architecture in accordance with another embodiment of the disclosed subject matter.

[0078] In FIG. Figure 20 shows the genetic sequence analysis pipeline.

[0079] In FIG. Figure 21 shows the processing steps using the genetic sequence analysis hardware platform.

[0080] Where appropriate, like reference numerals denote like structures, features, or elements.

DETAILED DESCRIPTION

[0081] To address these and possibly other problems associated with currently available solutions, methods, systems, products, etc., corresponding to one or more embodiments of the present invention may, among other possible advantages, provide an analysis device sequences to run a sequence analysis pipeline based on genetic sequence data.

[0082] The following provides a detailed description of various embodiments of a sequencing platform, on-line sequence analysis, and a system for performing one or more tertiary processing protocols.

[0083] In the most general terms, the body is made up of cells, these cells form tissues, tissues form organs, organs form systems, and these systems function together to ensure the functioning of the body and maintain the life of the individual. Thus, these body cells are the building blocks of life. More specifically, every cell has a nucleus, and within the said nucleus of each cell are chromosomes. Chromosomes are formed by deoxyribonucleic acids, which have an organized but coiled double helix structure. DNA itself consists of two oppositely directed but complementary chains of nucleotides that form genes that encode proteins that give the cell its structure and influence the functioning of tissues and organs of the body, as well as regulating their functions. In general, proteins perform most of the work of cells to maintain normal body processes and functions.

[0084] Given the large number of components of the body and the complexity associated with how they interact with each other to support the various processes and functions of the body, there are many ways that the body can function incorrectly at any of these various levels of organization. For example, in one such embodiment, the disorder may be in the way a particular gene encodes a given protein, and the disorder depends on the protein, and the nature of the disorder may cause the process of transition to a death state to begin.

[0085] Accordingly, in the diagnosis, prevention and/or treatment of such pathological conditions, determining the genetic characteristics of a subject can be extremely useful. For example, the genetic characteristics of a subject, if known, such as his or her gene composition, may be used for diagnostic purposes and/or to determine whether the subject has or has the potential to develop any pathological condition, and therefore may be used for prevention purposes. Also, knowledge of a subject's genome can be useful in identifying potential therapeutic interventions, such as drugs that may or may not be used in prophylactic or therapeutic treatment regimens without causing harm to the patient. In various embodiments, knowledge of a subject's genome can also be used in determining the effectiveness of a drug, and/or problematic side effects of use of such drug can be predicted and/or identified. Potentially, this knowledge of a subject's genome could be used to create a synthetic "designer" drug, where such a drug is individually created and optimized according to the specific genetic characteristics of the subject. In particular, as one example, a designed protein or nucleotide sequence may be produced in accordance with the unique genetic characteristics of a subject so as to turn off or turn on the transcription of genes that produce proteins in either excess or insufficient quantities, and thus eliminate pathological conditions.

[0086] Consequently, in some examples, the purpose of processing bioinformatics data is to determine the individual genomes of people, and such determinations can be used in gene detection protocols, as well as for the purposes of preventive and/or therapeutic interventions, to improve the quality of life of each individual subject and the human race as a whole. In addition, knowledge of a subject's genome can be used, for example, in drug discovery and/or clinical trials to better predict which drug is likely to work in a subject, if any, and/or which drug is likely to work. cause harmful side effects by analyzing an individual's genome and/or the protein profile derived from it and comparing it with the predicted biological response to administration of such a drug.

[0087] Such genomics and bioinformatics data processing typically involves three well-developed, but typically used separately, information processing phases. The first phase involves DNA/RNA sequencing, where the subject's DNA/RNA is obtained and subjected to various tests, thereby obtaining the subject's genetic code converted into a machine-readable digital code, such as a FASTQ file. The second phase involves using the resulting digital genetic code of the subject to determine the subject's genetic characteristics, such as determining the nucleotide sequences of the subject's genes and/or a file indicating variants, such as how the subject's genome differs from one or more reference genomes. The third phase involves conducting one or more analyzes of the subject's genetic characteristics to obtain therapeutically useful information from them. In the order in which they appear, these phases may be called primary, secondary and tertiary data processing, respectively.

[0088] Previously, for example, in Phase I or during primary data processing, the genetic material must be preprocessed, for example, by determining the nucleotide sequence to obtain genome sequence data that can be used further. Sequencing nucleic acids such as deoxyribonucleic acid (DNA) and ribonucleic acid (RNA) is a fundamental part of biological research. This definition is necessary for a large number of tasks and is very often used in scientific research, as well as in medical development. For example, the fields of genomics and bioinformatics are linked through the application of information technology and computer science to genetics and/or molecular biology. In particular, bioinformatics methods, such as those described herein, can be used to generate, process and analyze various genomic data, such as those obtained from humans, to determine qualitative and quantitative information about this data, which can then be used by various specialists in the development of individual and/or global diagnostic, preventive and/or therapeutic methods to identify, prevent and/or at least improve a pathological condition and, therefore, improve the safety, quality and effectiveness of medical care for the individual and/or society.

[0089] Typically, a DNA/RNA analysis method such as genetic diagnosis involves hybridization and nucleic acid detection. For example, various typical hybridization and detection approaches include the following steps. For genetic analysis, a sample of RNA or DNA from the subject to be analyzed can be isolated and immobilized on a support, and a probe with a known genetic sequence, such as a disease marker, can be labeled and applied to the entire support. If a disease marker is present, binding occurs, such as hybridization, and because the probe has been labeled, the hybridization event may or may not be detected, indicating the presence or absence of the disease marker in the subject's sample. Alternatively, as discussed above, when the hybridization reaction occurs adjacent to a reaction layer, for example configured to detect a reagent and/or reaction byproduct used in a suitable field-effect transistor (FET) device, there is no need to employ a labeled probe.

[0090] Typically, to determine a nucleotide sequence, first an unknown nucleic acid sequence, such as a single-stranded DNA and/or RNA sequence of a subject to be identified, is isolated, amplified, and immobilized on a support. A nucleic acid of known sequence, labeled with an identifiable label, is then brought into contact with a nucleic acid of unknown sequence in the presence of a polymerase. When hybridization occurs, the labeled nucleic acid binds to its complementary base in an unknown sequence immobilized on the surface of the support. The binding event can then be detected, for example, by optical or electrical means. These steps are then repeated until the entire DNA sample has been completely sequenced.

[0091] Typically, these steps are performed manually or using an automated sequencer, such as a Next Gen Sequencer (NGS), which can produce thousands to millions of sequences simultaneously in the next generation sequencing process. In this case, as described herein, a label-free direct sequencing system for DNA and/or RNA is used, such as a system on a computer chip, such as a complementary metal oxide semiconductor (CMOS) chip, where various components or the entire sensor apparatus of the sequencer can be embodied within or otherwise associated with said semiconductor chip. Such a system, as provided herein, allows for seamless integration of primary, secondary and/or tertiary processing, for example, using the same semiconductor chipset.

[0092] More specifically, a typical sequencing procedure, regardless of the type of sequencing device used, involves obtaining a biological sample from a subject, such as by venipuncture, hair, etc., and processing the sample to isolate a genetic component from it. After isolation, where the genetic sample is DNA, said DNA can be denatured and its strands separated. Since RNA is already single-stranded, this step may not be necessary when processing RNA. The isolated DNA and/or RNA, or portions thereof, can then be amplified, for example by polymerase chain reaction (PCR), to create a library of replicated strands that are now ready to be sequenced and read, for example by an automated sequencer that is configured to read replicated chains, for example, by synthesis, and thus determining the sequences of nucleotides that make up DNA and/or RNA. Additionally, in various embodiments, such as creating a library of replicated and amplified strands, it may be useful to provide excess overlap when preprocessing a given portion of the DNA and/or RNA. Performing such redundant overlap, such as using PCR, may require increased resources and time for sample preparation and will therefore be more expensive, but this often increases the likelihood that the final result will be more accurate.

[0093] Once a library of replicated DNA/RNA strands has been created, they can be input into an automated sequencer, such as NGS, which can then read the strands, eg by synthesis, to determine their nucleotide sequence. For example, replicated single-stranded DNA or RNA can be attached to a glass bead and introduced into a test vessel, such as a matrix. All necessary components for the replication of its complementary chain, including labeled nucleotides, are also sequentially added to the indicated vessel. For example, all "A", "C", "G" and "T" that can be labeled are added either one at a time or all together if labeled to see which nucleotide binds at position "one" of the single-stranded DNA or RNA.

[0094] After each attachment step in the marked model, light, such as a laser, is applied to the matrix. If the composition fluoresces, an image is created indicating which nucleotide is bound at the corresponding location. In an unlabeled model, a binding event can be detected, for example, by a change in resistance in a gate, such as a solution channel, near the reaction layer where a glass bead containing replicated single-stranded DNA or RNA is located. More specifically, when nucleotides are added one at a time, if a binding event occurs, its indicative fluorescence or resistance change will be observed. If no binding event occurs, the test vessel can be washed and the procedure repeated until the corresponding one of the four nucleotides binds to its complement at the appropriate location and its indicative change in conditions is observed. When all four nucleotides are added simultaneously, each can be labeled with different fluorescent tags, and the nucleotide that binds its complement at the appropriate position can be determined, for example by its fluorescence color. This significantly speeds up the synthesis process.

[0095] After the binding event has occurred, the complex is washed and the synthesis steps are repeated for position two. For example, a labeled or otherwise labeled nucleotide "A" can be added to the reaction mixture to determine whether the complement at position "one" in the associated sequence template molecule is the complement for "A" and, if so, the labeled reagent " A" will bind to the template sequence having this complement and will therefore fluoresce, after which all samples will be washed to remove any excess nucleotide reagents. Where binding has occurred, the bound nucleotide is not washed away. This process will be repeated for all nucleotides for all positions until all redundant nucleic acid segments, such as reads, have been sequenced and data collected. Alternatively, when all four nucleotides are added simultaneously, each labeled with a different fluorescent indicator, only one nucleotide will bind to its complement at the specified position and the others will be washed out, so that after washing the vessel, the specified vessel can be illuminated with a laser, and can be it is determined which nucleotide is bound to its complement, for example, by the color of its fluorescence. However, when a CMOS FET sensor is used, as described below, a coupling event can be detected by a conductance change that occurs in the vicinity of a suitably configured gate or other response region.

[0096] In particular, due to the need to use optically detectable, eg, fluorescent, tags in the sequencing reactions performed, the required equipment to perform such high-throughput sequencing can be bulky, expensive, labor-intensive to use, and not portable. For this reason, this application proposes a new approach to direct label-free DNA and/or RNA sequencing. For example, while various embodiments provide improved methods for performing NGS processing, other embodiments provide improved methods and devices for nucleic acid sequencing and/or processing that do not necessarily include NGS. For example, in particular, the present application proposes a detection method based on the use of various electronic analytical devices. Such direct electron detection methods have several advantages over the conventional NGS platform.

[0097] More specifically, the sensor and/or sensing device disclosed herein may be embedded in the substrate itself, such as that used in a biosystem-on-a-chip device, such as a complementary metal oxide semiconductor "CMOS" device. In particular, when using a CMOS device in genetic detection, a signal representing a hybridization event, such as hybridization and/or nucleic acid sequencing, can be directly obtained and processed on the microchip itself. In this case, automatic recognition can be achieved in real time and at a lower cost than is currently possible to achieve using conventional NGS processing. Additionally, standard CMOS substrate devices can be used for such electronic detection, making the process simple, inexpensive, fast, and portable.

[0098] For example, for next generation sequencing to become widely used as a diagnostic tool in the healthcare industry, sequencing equipment must be mass produced with a high degree of quality, portability and cost-effectiveness. One way to achieve this is to translate the DNA/RNA sequencing process into a format that fully utilizes the manufacturing base created for computer chips, such as complementary metal oxide semiconductor (CMOS) fabrication, which is the pinnacle of modern large-scale, high-quality, low-cost production using high technology. To achieve this, ideally the entire sensor apparatus of the sequencer could be housed on a standard semiconductor chip, manufactured, for example, in the same integrated circuit manufacturing facilities that are used to produce logic and memory chips.

[0099] Accordingly, in another aspect of the disclosure, provided herein is a field-effect transistor (FET) that can be fabricated or otherwise associated with a CMOS chip that is configured for use in performing one or more DNA/RNA sequence hybridization reactions. Such a FET may include a gate, a channel portion connecting the source and drain electrodes, and an insulating barrier that may be configured to separate the gate from the channel. Optimal operation of such a field effect transistor depends on control of the channel conductance and thus control of the drain current, for example, by means of a voltage that can be applied between said gate and said source electrode.

[00100] For high speed applications and to improve sensor sensitivity, FETs presented herein can be operated to quickly respond to changes in gate voltage (VGS). However, this requires the use of short gates and fast moving carriers in this channel. In view of this, modern FET sensors, for example for use in nucleic acid hybridization reactions and/or sequencing reactions, are configured to have very thin channels vertically and/or horizontally to provide high carrier transfer rates as well as increased sensitivity and accuracy of the sensors, which gives existing sensors particular advantages when performing nucleic acid sequencing reactions. Therefore, the devices, systems and methods of use presented herein are ideal for performing analysis of genomic information and in applications such as nucleic acid sequencing and/or genetic diagnostics.

[00101] Therefore, one aspect of the present invention is a chemically sensitive transistor, such as a field-effect transistor (FET), which is intended for the analysis of biological or chemical materials, which solves many of the current problems associated with nucleic acid sequencing and genetic diagnostics. Such field effect transistors can be fabricated on a primary structure such as a wafer, for example a silicon wafer. In various embodiments, said primary structure may include one or more additional structures, for example, in a stacked configuration, such as a layer of insulating material. For example, the insulating material may be applied over the primary structure and may be an inorganic material such as a silicon oxide, such as silicon dioxide or silicon nitride, or an organic material such as a polyimide, BCB or other similar material.

[00102] The primary and secondary structures, for example including an insulating layer, may include an additional structure comprising one or more of a conductive source and/or a conductive drain separated from each other by space and embedded in the primary structure and/or insulator material and /or located in the plane of the upper surface of the insulator. In various embodiments, these structures may further include a processor or may otherwise be coupled to the processor, for example, to process generated data, such as data received from sensors. Accordingly, the structures may be configured as an integrated circuit or may otherwise include an integrated circuit such as those described herein and/or may be an ASIC circuit, a structured ASIC circuit, or an FPGA circuit.

[00103] In particular, these structures can be made in the form of a complementary metal oxide semiconductor (CMOS), which, in turn, can be made in the form of a chemically sensitive field effect transistor containing one or more of the following: a conductive source, a conductive drain, a channel or well and/or processor. For example, the field effect transistor may include a CMOS structure having an integrated circuit that is fabricated on a silicon wafer that further includes an insulator layer, the insulator layer including a source conductive electrode and a drain conductive electrode, for example, embedded therein, wherein the source and drain electrodes may be made of metal, for example, a Damascus copper source electrode and a Damascus copper source electrode. In various embodiments, these structures may include a surface, such as a top surface, which surface may include a channel, such that the surface and/or channel may be configured to extend from a source to a drain, thereby forming a reaction zone.

[00104] In certain cases, said surface and/or channel may include a one-dimensional transistor material, a two-dimensional transistor material, a three-dimensional transistor material, and/or the like. In various embodiments, a one-dimensional (1D) transistor material may be included, which may consist of a carbon nanotube or a semiconductor nanowire. In other cases, said chamber and/or channel consists of a one-dimensional transistor material containing one or more carbon nanotubes and/or semiconductor nanowires, such as a semiconductor nanowire sheet.

[00105] In particular, a two-dimensional (2D) transistor material may be included, such as a 2D material that may be one or two atoms thick and may be distributed in a plane. In such cases, the 2D material may include or otherwise be composed of such elemental 2D materials as graphene, graphene (an allotrope of carbon consisting of a lattice of benzene rings linked by acetylene bonds), borophene (an allotrope of boron), germanene (an allotrope of germanium) , germanium (another allotrope of germanium), silicene (an allotrope of silicon), stanene (an allotrope of tin), phosphorene (an allotrope of phosphorus, sometimes called black phosphorus), or monoatomic layers of metals such as palladium or rhodium; transition metal dichalcogenides (which contain one transition metal atom for every two chalcogen atoms), such as molybdenum disulfide (MoS ₂ , sometimes called molybdenite), tungsten diselenide (WSe ₂ ), tungsten disulfide (WS ₂ ) or others; MChenes (transition metal carbides and/or nitrides, typically of the formula _Mn + _1Xn , where M is a transition metal and X is carbon and/or nitrogen), such as Ti ₂ C, V ₂ C, Nb ₂ C , Ti ₃ C ₂ , Ti ₃ CN , Nb ₄ C ₃ or Ta ₄ C ₃ (moreover, MHenes can be terminated with O, OH or F to obtain semiconductors with a small band gap); or organometallic compounds such as Ni HITP (Ni ₃ (2,3,6,7,10,11-hexaminotriphenylene) ₂ or 2D supracrystals (supracrystals are supraatomic periodic structures, where the atoms that are usually located at the nodes of the structure are replaced by their symmetrical ones complexes).It should be noted that transition metal dichalcogenides can contain one atom of any transition metal (Sc, Ti, V, Cr, Mn, Fe, Co, Ni, Cu, Zn, Y, Zr, Nb, Mo, Tc, Ru, Rh, Pd, Ag, Cd, Hf, Ta, W, Re, Os, Ir, Pt, Au, Hg, Rt, Db, Sg, Bh, Mt, Ds or Rg) paired with two atoms of any of the chalcogenides (S , Se or Te). In particular, the 2D material may include one or more layers of graphene, silicone, molybdenum disulfide, black phosphorus and/or metal dichalcogenides. In various embodiments, the three-dimensional (3D) material may be deposited on the surface and/or channel may include a dielectric layer.

[00106] Additionally, in various embodiments, a reaction layer, such as an oxide layer layered or otherwise deposited on a 1D layer, a 2D layer, such as a graphene layer, or a 3D layer, may be located on the surface and/or channel. Such an oxide layer may be a layer of aluminum oxide or a silicon oxide such as silica. In various embodiments, a passivation layer may be located on the surface and/or channel and/or on the corresponding reaction layer, layered or otherwise applied to a 1D layer, a 2D layer, such as a graphene layer, or a 3D layer. layer.

[00107] In some embodiments, the primary and/or secondary structures may be fabricated or otherwise configured to include a chamber structure or a dimple structure in and/or on said surface. For example, the dimple structure may be located on a portion of a surface, such as an outer surface, of a primary and/or secondary structure. In some cases, the dimple structure may be formed on top of at least a portion of or include a material of the 1D layer, 2D layer, e.g., graphene, and/or 3D layer, and/or may further include reaction layers eg oxide layers, and/or passivation layers. In various embodiments, the chamber structure and/or bows may define an opening, such as an opening that provides access to the interior of the chamber, such as providing direct contact with a 1D layer, such as a carbon nanotube or nanowire, a 2D layer, such as graphene, surface and/or channel.

[00108] Accordingly, in various embodiments, the present invention relates to a biosensor. Said biosensor includes a CMOS structure, which may include a metal-containing inlet end, for example, Damascus copper, as well as a metal-containing outlet end, for example, Damascus copper, a 1D or 2D layer, for example, a graphene layer, a surface or a channel extending from the electrode source to the drain electrode, and a dimple or chamber structure that may be located on an outer surface portion of the layered 1D, 2D or 3D dimple structure. In such a case, the dimple structure can be configured to define an opening that allows direct contact with the nanotube, nanowire, and/or graphene dimple or chamber surface. In various embodiments, an oxide layer and/or a passivation layer may be located on the surfaces of the chamber. Therefore, in some cases, a chemically sensitive transistor, such as a field-effect transistor (FET), including one or more nano- or microwells may be provided.

[00109] In some embodiments, the chemically sensitive field effect transistor may include a plurality of wells and may be configured as an array, such as a sensor array. As such, the system may include an array of wells including one or more, for example, a plurality of sensors, where each of the sensors includes a chemically sensitive field effect transistor having a source, a drain, and a reaction surface or channel extending from the source to drain. Such a matrix or such matrices can be used, for example, to detect the presence and/or changes in concentration of various types of analytes in a wide range of chemical and/or biological processes, including DNA/RNA hybridization reactions and/or sequencing reactions. For example, devices and/or systems including such devices described herein can be used in a method for diagnosing a disease and/or analyzing biological or chemical materials, such as whole genome analysis, genome typing analysis, microarray analysis, panel analysis, exome assay, microbiome assay, and/or clinical assay, such as cancer assay, NIPT assay, and/or UCS assay.

[00110] In a specific embodiment, the FET may be a graphene array FET (gFET) as described herein and may be used to facilitate DNA/RNA sequencing and/or hybridization techniques based on monitoring changes in hydrogen ion concentration (pH), changes concentrations of other analytes and/or binding events associated with chemistry involved in DNA/RNA synthesis, such as in a controlled reaction chamber or gFET-based sensor well. For example, a chemically sensitive field-effect transistor may be configured as a CMOS biosensor and/or may be adapted to increase the measurement sensitivity and/or accuracy of the sensor and/or corresponding array(s), for example, by including one or more surfaces or wells having a surface onto which a 1D and/or 2D and/or 3D material, a dielectric or reaction layer, a passivation layer and/or the like is applied. For example, in a particular embodiment, a chemically sensitive graphene field effect transistor (gFET), such as a gFET having a CMOS structure, is provided, wherein said gFET sensor, such as a biosensor, may include an oxide layer and/or a passivation layer, such as a layer located on the surface of a well or chamber, to increase measurement sensitivity and/or accuracy of the sensor and/or associated matrices. The oxide layer, if present, may consist of alumina, silica, silica, and the like.

[00111] The system may further include one or more fluid components necessary to carry out the reaction, a circuit component necessary to run the reaction processes, and/or a computing component for control and/or processing. For example, a liquid component may be included when the liquid component is configured to control one or more reactant flows over the array and/or one or more chambers thereof. Specifically, in various embodiments, said system includes a plurality of reaction sites, such as surfaces or wells, which in turn include a plurality of sensors and/or a plurality of channels and further include one or more fluid sources, containing a liquid containing a plurality of reagents and/or analytes for delivery to one or more surfaces and/or wells for carrying out one or more reactions therein. In some cases, a mechanism for generating one or more electric and/or magnetic fields may also be included.

[00112] The system may further include a circuit component that may include a sample and hold circuit, an address decoder, a bias circuit, and/or at least one analog-to-digital converter. For example, a sample-and-hold circuit may be configured to store an analog voltage value that is to be applied to a selected column and/or row of a matrix of a device according to the present invention, for example, during a sensing interval. In addition, the address decoder may be configured to generate column and/or row select signals for any column and/or row of the matrix to access a sensor with a given address in said matrix. The bias circuit may be coupled to one or more surfaces and/or chambers of the array and may include a bias component that may be adapted to apply a sense voltage and/or bias voltage to selected chemically sensitive field effect transistors of the array, such as a gate electrode of said array. transistor. An analog-to-digital converter can be configured to convert an analog value to a digital value.

[00113] A computing component may also be included, for example, where said computing component includes one or more processors, such as a signal processor; a base calling module configured to determine one or more bases of one or more sequenced nucleic acid reads; a mapping module configured to generate one or more seeds from one or more sequenced data reads and to perform a mapping function on the one or more seeds and/or reads; an alignment module configured to perform an alignment function on one or more displayed reads; a sorting module configured to perform a sorting function on one or more matched and/or aligned reads; and/or a variant calling module configured to perform a variant calling function on one or more mapped, aligned and/or sorted reads. In specific embodiments, the base caller of the base caller may be configured to correct multiple signals, such as phase and signal loss, for key normalization and/or to generate multiple corrected base calls for each thread in each sensor to produce multiple sequence reads. In various embodiments, said device and/or system may include at least one reference electrode.

[00114] In particular, the system may be configured to carry out a sequencing reaction. In such a case, the FET sequencing device may include an array of sensors having one or more chemically sensitive field effect transistors coupled thereto. Such transistors may include a cascade transistor having one or more source electrodes, drain electrodes, and/or gate electrodes. In such a case, a source electrode of said transistor may be directly or indirectly coupled to a drain electrode of said chemically sensitive field effect transistor. In some cases, a one- or two-dimensional channel may be included that may extend from the source electrode to the drain electrode, for example, where the 1D channel material may be a carbon nanotube or nanowire, and the 2D channel material may consist of graphene, silicone, phosphorene, molybdenum disulfide and metal dichalcogenide. The device may be further configured to include a plurality of column and row lines connected to sensors in a sensor array. In such a case, each column in the plurality of columns may be directly or indirectly coupled to drain electrodes of said transistors, such as cascaded transistors, of a corresponding plurality of pixels in the array, and likewise each row of the plurality of rows may be directly or indirectly coupled to source electrodes of said transistors, for example, cascaded transistors corresponding to a plurality of sensors in the matrix.

[00115] In some cases, a plurality of source and drain electrodes may be used having multiple reaction surfaces, eg, channel elements disposed therebetween, where each channel element includes one-, two-, or three-dimensional material. In such a case, a plurality of first and/or second conductive layers may be coupled to the first and second source/drain electrodes of the chemically sensitive field effect transistors in respective columns and rows in the array. In addition, a control circuit may be provided that may be connected to a plurality of columns and rows, for example, to read information from a particular sensor connected to a particular column and/or a particular row. The circuit may also include a bias component that may be configured to apply a sense voltage to a selected row and/or to apply a bias voltage to the gate electrode of a transistor, such as a field effect transistor and/or cascade transistor of a particular sensor. In a specific embodiment, the bias circuit may be coupled to one or more array cameras and may be configured to apply a read bias to selected chemically sensitive FETs via a conductive column and/or row. In particular, the bias circuit may be configured to apply a sense voltage to a particular row and/or to apply a bias voltage to the gate electrode of a transistor, such as a cascade transistor, during a sense interval.

[00116] A sensing circuit may be added and coupled to the matrix for sensing the charge associated with one or more gate configurations of the selected chemically sensitive field-effect transistor. The sensing circuitry may also be configured to read a specific sensor based on the sampled voltage level in a particular row and/or column row. In such a case, the read circuit may include one or more pre-charge circuits, for example, to pre-charge a particular column to a pre-charge voltage level before the read interval; and a sampling circuit, for example, to select a voltage level at the drain electrode of a selected transistor, such as a cascade transistor, during a sensing interval. The sampling circuit may also be included and include a sampling and holding circuit configured to store an analog voltage value on a particular column during a sensing interval, and may further include an analog-to-digital converter to convert the analog value to a digital value.

[00117] In another aspect, these 1D, 2D or 3D FET integrated circuits, such as gFETs, sensors and/or arrays according to the present invention, can be manufactured, for example, using any suitable complementary metal oxide semiconductor (CMOS) processing method, known from the prior art. In certain cases, such CMOS processing technology can be configured to increase the measurement sensitivity and/or accuracy of the sensor and/or sensor, while at the same time producing a significantly smaller sensor and dense gFET camera sensor area. In particular, the advanced fabrication technologies described herein, using 1D, 2D, 3D and/or oxide as the reaction layer, enable rapid acquisition of data from small sensors in large, dense sensor arrays. In particular embodiments, when an ion selective permeable membrane is used, the membrane layer may include a polymer such as perfluorosulfone material, perfluorocarbon material, PEEK, PBI, Nafion and/or PTFE. In some embodiments, the ion selective permeable membrane may include an inorganic material such as an oxide or glass. One or more different layers, such as reactive, passivation, and/or permeable membrane layers, may be fabricated or otherwise deposited by spin coating, anodizing, PVD, and/or sol-gel.

[00118] Accordingly, the CMOS field effect transistor device described herein can be used for sequencing a nucleic acid sample, where the nucleic acid sample serves as a template for the synthesis and sequencing of DNA/RNA that may be associated with or may be located close to the surface of the reaction zone , for example, a surface coated with graphene. Once immobilized, the template sequence can be sequenced and/or analyzed by performing one or more of the following steps. For example, a primer and/or polymerase, such as DNA and/or RNA polymerase, and/or one or more substrates, such as the deoxynucleotide triphosphates dATP, dGTP, dCTP and dTTP, may be added sequentially to the reaction chamber, for example after the initiation of the hybridization reaction to induce an elongation response. Once the corresponding substrate hybridizes with its complement in the template sequence, there is a concomitant change in the individual electrical characteristic voltage, such as the source-drain voltage (Vsd), measured as a result of a new local gate effect. If a reaction layer is added, such as an oxide layer deposited on a 1D, 2-D or 3-D surface, the sensitivity with which a binding event is detected can be enhanced, for example when the reaction layer is configured to receive and/or monitor changes in concentration hydrogen ions (pH), changes in the concentration of other analyzed compounds.

[00119] Therefore, for each extension reaction with a corresponding, for example, complementary substrate, there will be a change in characteristic voltage and/or pH concentration. For example, as described herein, a field device for nucleic acid sequencing and/or gene detection may be located in a sample chamber or well of a flow cell, and a sample solution, for example, containing a polymerase and one or more substrates, such as nucleic acids, may be introduced into the sample solution chamber, for example, through one or more fluid components of the system. In various embodiments, the reference electrode may be located upstream, downstream, or in fluid contact with the field device, and/or the source and/or drain electrodes may themselves serve as electrodes used to detect hybridization, and the voltage to the shutter can be supplied when needed.

[00120] In particular, in the typical extension reaction described above, polynucleotides are synthesized if the added substrate is complementary to the base sequence of the target DNA/RNA template and/or primer template. If the added substrate is not complementary to the next available base sequence in the template, hybridization and extension will not occur. Because nucleic acids such as DNA and RNA have a negative charge in aqueous solutions, hybridization leading to extension can be consistently detected by changing the charge density at the reaction surface and/or in the reaction chamber. Such detection can be improved by the ability to detect an increase in ion concentration, for example by detecting a change in pH. Because substrates are added sequentially, it is easy to determine which nucleotide is bound to the template, thereby facilitating the extension reaction. Accordingly, as a result of elongation, the negative charge on the surface of the graphene layer gate, the surface of the insulating film, and/or the surface of the side wall of the reaction chamber will be increased. Such an increase can then be detected, for example, by changing the source gate voltage and/or ion concentration, as described in detail herein. By identifying the addition of the substrate that caused the signal or pH change in the gate-source voltage, the base sequence of the target nucleic acid can be determined and/or analyzed.

[00121] In particular, regardless of the sequencing device used, such as an NGS and/or FET-based sequencing device as described herein, this iterative synthesis process continues until the entire DNA/RNA template strand has been replicated in the indicated container. Typically, the typical length of sequence replicated in this manner is from about 100 to about 500 base pairs, such as from 150 to 400 base pairs, including from about 200 to about 350 base pairs, such as from about 250 base pairs to about 300 base pairs , depending on the sequencing protocol used. In addition, the nucleotide length of these template segments can be predetermined, for example designed, to suit any particular sequencing engine and/or protocol under which it is performed.

[00122] The end result is a read or read that consists of a replicated segment of DNA/RNA, e.g., about 100 to about 1000 nucleotides or more in length, that has either been labeled such that each nucleotide in the sequence, e.g., read, is known based on its label, or determined and known by a change in gate characteristic, such as a change in voltage and/or pH. Therefore, since the human genome consists of approximately 3.2 billion base pairs, and various known sequencing protocols typically result in tagged replicated sequences, e.g., reads ranging in length from about 100 or 101 bases to about 250, or about 300, or about 400 bases, the total number of resulting segments that need to be sequenced, and therefore the total number of reads generated, can be from about 10,000,000 to about 40,000,000, such as from about 15,000,000 to about 30,000,000, depending on how long the replicated tagged sequences are. Therefore, a sequencer can typically generate about 30,000,000 reads, for example when the read length is 100 nucleotides in length, to cover a specified genome once. However, as discussed herein, due to the compressed nature of the present sequencing chip format provided herein, much larger read lengths such as 800 bases, 1000 bases, 2500 bases, 5000 bases, up to 10,000 bases.

[00123] Additionally, as noted above, such procedures may benefit from an excess set of DNA/RNA reads of about 5 times, or about 10 times, or about 20 times, or about 25 times, or about 30 times, or about 40 times, or about 50 times, or about 100 times, or about 200 times, or about 250 times, or about 500 times, or about 1000 times, or about 5000 times , or even about 10,000 times or more, and thus the amount of primary processing that needs to be done and the time spent doing so can be quite significant. For example, with a 40-fold information excess, where the different synthesized reads are designed to overlap to some extent, up to 1.2 billion reads may need to be synthesized. Typically, the vast majority, if not all, of these tagged sequences can be generated in parallel. The end result is that the original biological genetic material is processed, for example, using a sequencing protocol such as those summarized herein, and a digital representation of that data is generated, which digital representation of the data may be subjected to a downstream processing protocol.

[00124] In particular, the genetic material of a subject may be replicated and sequenced such that a measurable electrical, chemical, radioactive and/or optical signal is generated, which is then converted, for example, by a sequencer and/or processing device associated therewith, into a digital representation of the subject's genetic code. More specifically, primary processing may include converting images, such as recorded flashes of light or other electrical or chemical signal data, into FASTQ file data. Accordingly, this information is stored as a FASTQ file, which can then be sent for further processing, such as secondary processing. A typical FASTQ file includes a large set of reads representing digitally encoded nucleotide sequences in which each predicted base in the sequence has been identified and given a probability value that the specified base at a specified position is incorrect.

[00125] In many cases, it may be useful to further process the digitally encoded sequences obtained from the sequencer and/or sequencing protocol, for example, by performing secondary processing of the digitally represented data. This secondary processing, for example, may be necessary in order to assemble a complete genomic profile of an individual, for example where all of the individual's genetic factors are determined, where the sequence position of each nucleotide of each chromosome is determined so that the composition of the entire genome of the individual is identified. In such processing, an individual's genome may be assembled, for example, by comparison to a reference gene, such as a standard, such as one or more genomes obtained from the Human Genome Project, to determine how the individual's genetic characteristics differ from the reference genome(s). This process is commonly known as option identification. Since the difference between DNA/RNA of different individuals ranges from 1 to 1000 base pairs, this process of identifying variants can be very labor-intensive and time-consuming.

[00126] Accordingly, in a typical secondary processing protocol, the genetic makeup of a subject is collected through comparison to a reference genome. This comparison involves reconstructing an individual's genome from millions and millions of short sequence reads and/or comparing the individual's entire DNA and/or RNA to an example DNA and/or RNA sequence model. In a typical secondary processing protocol, a FASTQ file containing raw sequenced read data is received from the sequencer. For example, in some cases there may be up to 30,000,000 or more reads spanning a subject's genome, assuming there is no oversampling, such as when each read is approximately 100 nucleotides in length. Therefore, in such an example, to compare a subject's DNA/RNA genome to a standard reference genome, it is necessary to determine where each of these reads maps onto the reference genome, e.g., how each one aligns relative to the other and/or how each read might also be sorted by chromosome order to determine what position each read is in and which chromosome it belongs to. One or more of these functions may precede the execution of a variant detection function on the full-length sequence. Once it is determined which part of the genome each read belongs to, the full-length genetic sequence can be determined, and the differences between the subject's genetic code and the genetic reference code can then be assessed.

[00127] Since the human genome is over 3 billion base pairs in length, efficient automated sequencing protocols and equipment have been developed to perform sequencing of such DNA/RNA genomes over periods of time that can be clinically useful. Thus, innovations in automated sequencing have made it possible to sequence an entire genome in a matter of hours or days, depending on the number of genomes sequenced, taking into account the oversampling of samples, and the amount of processing resources involved in the work. Therefore, given this advance in generation, a large amount of sequenced data can be generated in a relatively short period of time. The result of this progress, however, is the development of a limiting factor (“bottleneck”) at the secondary processing stage. In an attempt to help overcome these limiting factors, various software-based algorithms, such as those described herein, have been developed to help speed up the process of assembling sequenced DNA and/or RNA from a subject, for example through a reference-based assembly process.

[00128] For example, reference-based assembly in a typical secondary processing assembly protocol involves comparing a subject's sequenced genomic DNA and/or RNA to sequenced genomic DNA and/or RNA of one or more standards, such as known reference sequences. Various algorithms have been developed to help speed up this process. Typically, these algorithms involve some version of one or more of: mapping, aligning, and/or sorting millions of reads obtained from digital files, such as FASTQ files, that are fed by a sequencer to determine which location on each chromosome belongs to or otherwise Each specific read is located there. Often a common feature in the operation of these various algorithms is their use of an index and/or array to speed up their processing function.

[00129] For example, with respect to mapping, a large number of sequenced reads, for example all of them, can be processed to determine possible locations in the reference genome to which these reads could be aligned. One method that can be used for this purpose is to directly compare the read to a reference genome to find all match positions. Another method is to use a prefix or suffix array or construct a prefix or suffix tree to map reads to different positions in a DNA/RNA reference genome. A typical algorithm useful in performing this function is the Burrows-Wheeler transform, which is used to map reads to a reference using a compression formula that compresses repeated sequences of data.

[00130] Another method is to use a hash table, for example, where a selected subset of reads, k-mers of selected length "k", such as a seed, are placed in the hash table as keys, and the reference sequence is split into parts equal to is a k-measure in length, and these parts and their locations are inserted by an algorithm into the hash table at those places in the table to which they are mapped according to the hash function. A typical algorithm for performing this function is BLAST, the Basic Local Alignment Search Tool. Such hash table-based programs compare the nucleotide or protein sequences of interest to one or more standard reference sequence databases and calculate the statistical significance of the matches. In this way, the likely location of any given read relative to the reference genome can be determined. These algorithms are useful because they require less memory, transformation, and therefore require less computational resources and time to perform their functions than would otherwise be the case, for example, if a subject's genome were assembled by direct comparison, for example, without the use of these algorithms.

[00131] Additionally, an alignment function can be performed to determine all possible mapping locations of a given read onto the genome, for example in cases where a read can be mapped to multiple positions in the genome that are in fact the location from which it was actually obtained. for example, by sequencing from that location using the original sequencing protocol. This function can be performed on a number of genome reads and can produce a string of ordered nucleotide bases representing part or all of the subject's genomic DNA and/or RNA sequence. Along with the ordered genetic sequence, each nucleotide position can be assigned a score representing, for any given nucleotide position, the probability that the nucleotide, such as "A", "C", "G", "T" (or "U"), predicted to this position is indeed the nucleotide that belongs to this designated position. Typical algorithms for performing the alignment function are the Needleman-Wunsch and Smith-Waterman algorithms. In either case, these algorithms perform sequence alignments between a subject's DNA and/or RNA genomic sequence string and a reference genomic sequence string, thereby comparing selected segments of possible lengths instead of comparing entire genome sequences to each other.

[00132] Once reads have been assigned positions, for example, relative to a reference genome, which may include determining whether the read belongs to a particular chromosome and/or is offset from the start of that chromosome, the reads can be sorted by position. This may allow subsequent analyzes to take advantage of the oversampling described herein. All reads that overlap a given position in the genome will be adjacent to each other after sorting and can be organized into a pileup and easily examined to determine whether most of them agree with the reference value or not. If not, the option can be flagged.

[00133] Although data and other algorithms like them address the limitations inherent in secondary processing in their own way, faster runtimes and greater accuracy are still needed. In particular, although there is progress in the generation of raw data, such as DNA/RNA sequence data generated, the development of information technology does not stand still, which leads to limiting factors in data analysis. These limiting factors have been reduced to some extent by the development of various algorithms, such as those described above, which help speed up this analysis, but there is still a need for new technologies to perform data generation, calculation, storage and/or analysis of such data. , this particularly applies to genomic sequence analysis, such as the secondary processing step.

[00134] For example, using standard next-generation technologies, sequencing the human genome can take several hours, up to a day, and when using standard protocols to perform secondary processing of such resulting sequenced genomic data, processing of the sequenced data can take up to three (3) days or even up to a week or more to generate clinically relevant information about an individual's genomic sequence. By using various optimized devices, algorithms, methods and/or systems, the processing time from primary to secondary can be reduced to as little as 27 to 48 hours. However, achieving such rapid results typically requires essentially simultaneous and parallel processing of all generated reads, for example 30 million reads of 100 nucleotides each. Such parallel processing requires expensive processing power, including massive CPU resources, and still takes a relatively long time.

[00135] In addition, in various examples, increased accuracy of the results is necessary. This increased accuracy can be achieved by providing some oversampling of the genome being sequenced. For example, as described above, it may be necessary to process a subject's DNA such that at any given location in the nucleotide sequence there is an oversampling of that region. As stated above, oversampling may be required in any given region of the genome, including 10X, 15X, 20X, 25X, 30X, 40X, 50X, 100X, 250X, or even 500X or 1000X times or more. However, when there is an oversampling of the genome, say 40X times, the number of reads that need to be processed is approximately 30 million x 40 (depending on the length of the reads), resulting in 1.2 billion reads needing to be processed when there are oversampling the genome by a factor of 40X. Therefore, although such oversampling typically results in greater accuracy, it comes at the cost of more time and requires larger processing resources since each genomic region is covered 1 to 40 times. Moreover, in some oncology applications in which a clinician is trying to distinguish the mutated genome of cancer cells in the bloodstream from the genome of healthy cells, oversampling by a factor of 500X, 1000X, 5000X, or even 10,000X may be used.

[00136] The present invention is therefore directed to such new technologies that can be implemented in a single genomic and/or bioinformatics protocol, such as a pipeline, or a sequence thereof, to perform genetic acquisition and/or analysis, such as primary and/or or secondary processing of the obtained genomic sequence data or part thereof. Sequence data can be obtained directly from high-throughput automated sequencing systems, such as ROCHE's 454 Sequencing by Synthesis automated sequencer, ILLUMINA's HiSeq x Ten or Solexia automated sequencers, Sequencing by Ligation and Discovery sequencers oligonucleotide" (“Sequencing by Oligonucleotide Ligation and Detection” (SOLiD)) or Ion Torrent sequencing from LIFE TECHNOLOGIES and/or the “Single Molecule Fluorescent Sequencing” sequencer from HELICOS GENETIC ANALYSIS SYSTEMS, or the like, for example, through a direct connection to a sequencing processing unit, or the sequenced data can be obtained directly, for example, by sequencing on a chip configuration, such as a graphene layered FET sensor containing a complementary metal oxide semiconductor (CMOS) sequencing chip , as described in this document. Such sequenced data may also be obtained remotely, for example from a database, such as over the Internet, or from other remote data accessible via wireless communication protocols, such as WiFi, Bluetooth or the like.

[00137] In accordance with certain aspects, these genetic acquisition and/or analysis technologies may use advanced algorithms that may be implemented in software that are performed with less processing intensity and/or less time and/or a higher percentage of accuracy . For example, in some embodiments, improved devices and methods are provided for producing genetic sequence information, such as in primary processing protocols as disclosed herein, and/or improved algorithms for performing secondary processing on it, as disclosed herein. In various specific embodiments, improved devices, systems, methods of using them, and applied algorithms are aimed at more efficiently and/or more accurately performing one or more of the functions of sequencing, mapping, alignment, and/or sorting, such as generating and/or analyzing a digital representation DNA/RNA sequence data obtained from a sequencing platform, for example, in FASTQ file format obtained from an automated sequencer and/or on-chip sequencer, such as one described above.

[00138] In addition, some embodiments provide improved algorithms to more efficiently and/or more accurately perform one or more of the functions of local realignment, duplicate flagging, ground truth recalibration, variant detection, compression, and/or decompression. In addition, as described in more detail below herein, in certain aspects, these genetic production and/or analysis technologies may employ one or more algorithms, such as advanced algorithms that may be implemented in hardware, that run less intensively. processing, and/or with less time and/or a higher percentage of accuracy than various software implementations for doing the same.

[00139] In specific embodiments, a technology platform is provided for DNA/RNA sequencing to obtain genetic sequence data and/or to perform genetic analyses, wherein the platform may include performing one or more of the functions of: sequencing, mapping, alignment, sorting, local realignment, duplicate marking, base quality recalibration, variant identification, compression and/or decompression, and/or may also include tertiary processing protocols as described herein. In certain cases, the implementation of one or more of these platform functions is designed to generate and/or perform one or more of determining and/or reconstructing a consensus genomic sequence of a subject, comparing the genomic sequence of the subject with a reference sequence, e.g., a reference or model genetic sequence, determining whether how a subject's genomic DNA and/or RNA differs from a reference, e.g., to identify variants, and/or to perform tertiary analysis on the subject's genomic sequence, e.g., to analyze a whole genome, e.g., to analyze genome-wide variation, and/or to genome typing analysis, gene functional analysis, protein functional analysis, e.g. protein binding assay, numerical and/or assembled genome and/or transcriptome analysis, micropanel analysis, panel analysis, exome analysis, microbiome analysis and/or clinical analysis, e.g. cancer analysis , non-invasive prenatal testing (NIPT) assay and/or UCS assay, as well as for various diagnostic and/or prophylactic and/or therapeutic evaluation assays.

[00140] In particular, after genetic data has been generated and/or processed, for example, in one or more primary and/or secondary processing protocols, for example, mapped, aligned and/or sorted, for example, to create one or more files determining variants, for example, to determine how a subject's genetic sequence data differs from one or more reference sequences; in another aspect, the present invention may relate to performing one or more other analytical functions on generated and/or processed genetic data, for example, to further processing such as tertiary processing. For example, the system may be configured to perform further processing on generated and/or secondary processed data, such as by running it through one or more tertiary processing pipelines, such as one or more of a genome pipeline, an epigenome pipeline, a metagenome pipeline, a shared genotyping, MuTect2 pipeline or other tertiary processing pipeline, for example through the devices and methods disclosed herein. For example, in various examples, an additional level of processing may be provided, for example, for disease diagnosis, therapeutic treatment and/or prophylactic prevention, including, for example, non-invasive prenatal testing (NIPT), neonatal intensive care unit (NICU), cancer, performed in the laboratory research (LDT), agrobiology (AgBio) and other types of diagnosis, prevention and/or treatment of such diseases that use data generated by one or more of the specified primary, secondary and/or tertiary pipelines. Therefore, the devices and methods described herein can be used to generate genetic sequence data, which can then be used to generate one or more variant definition files and/or other associated information that can be further processed by other tertiary pipelines. processing in accordance with the devices and methods described herein, for example, for the diagnosis of specific and/or general diseases, as well as for preventive and/or therapeutic measures and/or developmental methods.

[00141] Additionally, in various embodiments, the bioinformatics processing mode disclosed herein can be used to create one or more masks, a genomic reference mask, a standard mask, a disease mask, and/or an iterative feedback mask that can be added to a mapper and/or aligner, for example together with a reference, wherein the set of masks is configured to identify a particular area or object of interest. For example, in one embodiment, the methods and apparatus disclosed herein can be used to create a genomic reference mask, for example by creating a set of masks that can be loaded into a mapper and/or aligner along with the reference, wherein the set of masks is identifiable particularly important and/or relevant areas, for example for a specialist and/or subject, and/or to identify areas with increased sensitivity to errors. In various embodiments, a set of masks can provide intelligent direction to the mapper and/or aligner, such as which regions of the genome to focus on for improved quality. In this way, masks can be created in a layered manner to provide variable levels or iterations of direction based on different specific use cases. Accordingly, each mask can identify regions of interest and provide a minimum quality target for the region. In addition, a standard mask can be used to provide targeting, for example, to identified, eg typical, "particularly important" regions of the genome. Such regions may include known coded regions, monitored regions, etc., as well as regions that are well known to produce errors. In addition, a disease mask, or application-specific mask, can be used in a set of masks that identifies areas of particular interest, such as areas that require high levels of accuracy based on known markers, such as cancer. In addition, iterative feedback masking can be applied, for example by adding a new, case-specific mask that can be specifically designed using feedback from a tertiary analysis system (like Cypher Genomics) that identifies regions of interest based on observed errors or inconsistencies.

[00142] As stated above, in one aspect, one or more of these platform functions, e.g., mapping, alignment, sorting, re-alignment, duplicate marking, ground quality recalibration, variant detection, one or more tertiary processing, compression, and /or unpacking, are designed to be implemented in software. In another embodiment, one or more of these platform functions, e.g., mapping, alignment, sorting, local realignment, duplicate marking, ground quality recalibration, decompression, variant detection, tertiary processing, compression, and/or decompression functions, are configured to implementation in hardware.

[00143] Accordingly, in certain cases, methods are provided herein, where the methods include executing an algorithm, such as an algorithm for implementing one or more genetic analysis functions such as mapping, alignment, sorting, realignment, duplicate flagging, qualitative recalibration basis evaluations, variant determination, compression and/or decompression, where the algorithm is optimized according to the way it is to be implemented. In particular, when an algorithm is to be implemented in a software solution, the algorithm and/or its supporting processes are optimized so that they run faster and/or with higher accuracy when executed by that environment. Likewise, when the functions of an algorithm must be implemented in a hardware solution, the hardware is designed to perform those functions and/or the processes that serve them in an optimal manner to perform faster and/or with higher accuracy when executed by that environment. These methods, for example, can be used, for example, in an iterative variant determination procedure.

[00144] Therefore, in one aspect, systems, devices and methods are provided herein for implementing bioinformatics protocols, for example, for performing one or more genetic data analysis functions, such as genomic data, for example, through one or more optimized algorithms and/or one or more optimized integrated circuits, for example on one or more hardware processing platforms. Thus, in one case, systems and methods are provided for implementing one or more algorithms for performing one or more steps of analyzing genomic data in bioinformatics protocols, for example, where the steps may include performing one or more of: mapping, alignment, sorting, local realignment, marking of duplicates, recalibration of qualitative assessment of the base, identification of variants, compression and/or decompression. Alternatively, systems and methods are provided for implementing functions of one or more algorithms for performing one or more steps of genomic data analysis in a bioinformatics protocol as defined herein, wherein the functions are implemented on a hardware accelerator that may or may not be coupled to one or more more general purpose processors and/or supercomputers.

[00145] More specifically, in some cases, methods have been proposed for performing secondary analytics on data related to the genetic makeup of a subject. In one case, the analytics to be performed may include a reference-based reconstruction of the subject's genome. For example, reference-based mapping involves the use of a reference genome, which may be generated by sequencing the genome of one or more individuals, or it may be a pool of DNA from different individuals that is combined in such a way as to create a prototypic standard reference genome with which the DNA of any individual can be compared, for example, to determine and reconstruct the individual's genetic sequence and/or to determine the difference between their genetic makeup and this standard reference, for example, to identify variants.

[00146] In particular, the reason for performing secondary analysis on a subject's sequenced DNA is to determine how the subject's DNA differs from the reference DNA. More specifically, to determine one, many, or all differences between a subject's nucleotide sequence and a reference nucleotide sequence. For example, differences between the genetic sequences of any two randomly selected individuals occur 1 time per 1000 base pairs, which, given the over 3 billion base pairs in the entire genome, amounts to a variation of up to 3,000,000 different base pairs per person. Determining these differences may be useful, for example, in a tertiary analysis protocol, for example, to predict the possibility of a disease state, for example, due to a genetic disorder, and/or the likelihood of success of a prophylactic or therapeutic intervention, for example, based on how the prevention or treatment is expected to interact therapy with the subject's DNA or proteins formed in this process. In various cases, it may be useful to perform a reconstruction of a subject's genome, either for the first time or from a reference, to reinforce the results of one with the results of the other, and to improve the accuracy of the variant identification protocol, if required.

[00147] In various examples, as indicated above, when performing a primary processing protocol, it may be useful to provide oversampling for one or more regions of a subject's genome. These regions may be selected based on known regions of increased variation, regions of suspected increased variation, for example, based on a subject's condition and/or the entire genome. In its basic form, as stated above, based on the type of sequencing protocols performed, sequencing produces reads, such as reads, which are a digital representation of a subject's genetic sequence code. These read lengths are typically generated based on the type of sequencing equipment used. For example, the ROCHE 454 automated sequencer typically produces reads ranging in length from 100 or 150 bp to approximately 1000 bp; For some ILLUMINA technologies, read lengths typically range from about 100 or 101 to about 150 base pairs and 250 base pairs for other technologies; for LIFE TECHNOLOGIES' SOLiD technology, read lengths are typically between approximately 50 and approximately 60 base pairs, and for Ion Torrent technology, between 35 and 450 base pairs; and for HELICOS GENETIC ANALYSIS SYSTEMS, the length of the reads may vary, but can typically be less than 1000 nucleotides.

[00148] However, since the processing of a DNA sample required to obtain reads of a given length of a specific size is labor-intensive as well as chemically expensive, and since sequencing itself is often dependent on the operation of the sequencing equipment, there is a possibility of errors occurring in the sequencing process , thereby introducing disturbances into the part of the sequenced genome in which the error occurred. Such errors can cause problems, especially when the purpose of reconstructing a subject's genome is to determine that it, or at least part of the genome, differs from the standard or reference model. For example, a design or chemistry error that results in the replacement of one nucleotide, for example in a read, by another will give a false indication of variation that actually does not exist. This may lead to incorrect determination of grounds and may also lead to false indication of a disease state or the like. Accordingly, due to the possibility of hardware, chemical, and/or even human error in performing a sequencing protocol, in many cases it is desirable to provide redundancy in the analysis system, for example by providing redundancy of samples throughout the genome. In particular, when providing an automated sequencer with a FASTQ file defining the sequence of reads containing nucleotides at given positions, coupled with the likelihood that the assignment of a given nucleotide definition to a particular position is actually false, such as a base definition, it is often necessary to use techniques such as redundancy sampling of samples to ensure that base determinations made during the sequencing process can be detected and corrected.

[00149] Therefore, when performing the methods described herein, in some cases, the primary processing protocol is performed in such a way as to obtain a sequenced genome in which a portion or all of the genome is oversampled by approximately 10X, 15X, 20X, 25X, 30X, 40X, 50X times or more. Accordingly, when the length of the reads is approximately 50-60 base pairs, such oversampling may result in the presence of approximately 2 to approximately 2.5 billion reads, or when the length of the reads is approximately 100 or 101 read pairs, oversampling may result in having about 1 to about 1.2 billion reads, and when the length of the reads is about 1000 base pairs, about 50 to about 100 million reads can be generated by the sequencer, for example when the oversampling is about 40X times. Specifically, in this example, due to oversampling by a factor of 40X, at any given location in the genome there are expected to be 40 reads covering any position, even though a given position may be at the beginning of one read, the middle of another, and the end of another. one, however it is expected to be covered approximately 40 times.

[00150] Thus, such oversampling provides regions of the sequenced genome that are covered by many reads, eg, duplicates, eg up to 40 reads, eg when oversampling is approximately 40X times. These at least partial duplicates are useful in determining whether any given variation in any particular read is in fact a true genomic variation or merely an artifact of equipment or chemistry. Thus, oversampling can be used to improve accuracy in reconstructing a subject's genome, especially in cases where the subject's genome needs to be compared to a reference genome to determine when the subject's genetic sequence differs from the reference genetic sequence. Thus, as described in more detail later in this document, it can be confirmed that any given variation between the reconstructed sequence and the model is in fact caused by the presence of a real variant and not by an error in the initial processing of the DNA sample, read alignment software, etc.

[00151] For example, when constructing the genetic sequence of a subject's sequenced DNA, it is necessary to determine where which nucleotide should be in a growing string of nucleotides. To determine which nucleotide should be where, different reads can be organized and a cluster of reads covering duplicate positions can be obtained. This allows comparison of all reads covering the same positions to more accurately determine whether there is a valid variant at any given position, or whether there is an error in any of the reads at the position in question in the cluster. For example, if there are only one or two reads out of 40 in which a particular nucleotide is at position X, and all 38 or 39 other reads agree that another nucleotide is at that position, then the two outlier reads can be excluded as erroneous, at least in the specified specific position.

[00152] In particular, when there are many reads generated for any location in a subject's genome, it is most likely that for any given nucleotide position there will be many overlaps or clusters. These clusters cover any specific location and can be useful in determining the correct genome sequence of a subject with greater accuracy. For example, as stated above, sequencing produces reads and, in various examples, the resulting reads are subject to oversampling such that at different positions different specific reads will overlap. This overlap is useful for determining the actual genome of a sample, for example with a high probability that it will be correct.

[00153] The objective may thus be to incrementally scan the reference genome multiple times, as described in more detail later herein, to more accurately reconstruct the subject's genome, and optionally determine how the subject's genome differs from another genome, such as a model genome ,The use of clusters can identify errors with greater ,precision, such as chemical, equipment, or read errors, and ,distinguish them from valid variants. In particular, when a subject has a valid variation in position X, the majority of reads in the cluster should confirm, i.e. contain the specified variation. Statistical analysis procedures, such as those disclosed herein, may then be performed to determine the subject's actual genetic sequence with all its variants relative to the reference genome.

[00154] For example, when the genetic sequence of a subject needs to be rearranged relative to the use of a reference genome, after generating reads, such as a read aggregation, the next steps may be to map, align, and/or sort the reads relative to one or more reference genomes (e.g., the more taken as examples of reference genomes are available as models, the better the analysis should be) and thus the reconstruction of the subject's genome, resulting in sets of reads being mapped and/or aligned with respect to the reference genome(s) at all possible positions along the chain in which there is a match, and in each such position they are assigned a probability score reflecting the likelihood that they actually belong to that position.

[00155] Accordingly, in various examples, after reads are generated, their positions are mapped, for example, potential reference genome locations to which the reads can be mapped and their sequential order aligned can be determined, the actual genome sequence of the subject can be determined, for example due to performing the function of sorting the aligned data. Additionally, once the actual genome of a sample is known and compared to a reference genome, variations between the two genomes can be determined, and a list of all variations/deviations between the reference genome and the sample genome can be determined and recalled. Such variations between two genetic sequences can be due to a number of reasons.

[00156] For example, a single nucleotide polymorphism (SNP) is possible, for example where one base in a subject's genetic sequence has been replaced by another; more extensive substitutions of multiple nucleotides are possible; an insertion or deletion is possible, such as when one or more bases are added to or removed from a subject's genetic sequence; and/or there may be a structural variant, such as one caused by the crossing of two chromosome legs, and/or there may simply be a misalignment resulting in a shift in sequence. In various examples, a variant definition file may be generated containing all variations in a subject's genetic sequence relative to reference sequences. Specifically, in various embodiments, the methods of the present invention may include generating a variant definition file (VCF) identifying one or more, e.g., all, genetic variants in an individual whose DNA has been sequenced, e.g., according to one or more reference genomes. A VCF file in its basic form is a list of variation positions and their type: for example, chromosome 3, at position X, “A” replaced by “T,” etc.

[00157] However, as stated above, to generate such a file, the subject's genome must be sequenced and rebuilt before its variants can be determined. However, there are several problems that may arise when attempting to form such an assembly. As stated above, there may be problems with the chemistry, sequencer, and/or human errors that occur during the sequencing process. In addition, there may be genetic artifacts that make such reconstruction problematic. For example, a challenge in performing such assemblies is that sometimes there are huge portions of the genome that repeat themselves, such as long sections of the genome that include the same strings of nucleotides. Therefore, since any genetic sequence is not unique everywhere, it may be difficult to determine where in the genome the identified read is actually mapped and aligned.

[00158] For example, depending on the sequencing protocol used, shorter or longer reads may be obtained. Longer reads are useful in that the longer the read, the less likely it is to appear in multiple places in the genome. Having to evaluate fewer possible positions can also speed up the system. However, the longer the reads, the more problematic they can be, as they are more likely to contain real or spurious variations, such as those due to an SNP, an indel (insertion or deletion) (InDel), a hardware error, or the like, resulting that there will be no matches between the read and the reference genome. On the other hand, shorter reads are beneficial because the shorter the read, the less likely it is to cover the position that encodes the variant. The problem with shorter reads, however, is that a shorter read is more likely to appear at multiple positions in the genome, requiring additional processing time and resources to determine which of all possible positions is most likely to be the actual alignment position . The ideal that can be achieved, for example by practicing the methods described herein, is that a variant definition file can be obtained that generates a list of the sequenced genome (the genome of interest) that shows where all the base pair variants are located, ensuring that that each variant identified is a valid variant and not simply a chemical error, an equipment error, or a human-induced read error.

[00159] Thus, there are two main possibilities for variation. First, there is actual variation at that particular site of interest, for example when the human genome at a particular site is truly different from the reference, for example there is natural variation due to a SNP (single base substitution), insertion or deletion (one or more nucleotides long ), and/or there is a structural variant, such as when DNA material from one chromosome crosses onto another chromosome or stem, or when a certain region occurs twice in the DNA. Alternatively, the variation may be due to a problem in the read data due to chemistry or equipment error, sequencer or aligner error, or other human error. Accordingly, the methods described herein can be used to compensate for these types of errors and, in particular, to distinguish errors in variation due to chemistry, equipment, or humans from actual variation in the sequenced genome. More specifically, the methods, apparatus, and systems for implementing them described herein are designed to clearly distinguish between these two different types of variations and, therefore, to better ensure the accuracy of any generated variation files to correctly identify true variations.

[00160] Additionally, in various embodiments, once a subject's genome has been reconstructed and/or a VCF file has been generated, such data may then be subjected to tertiary processing for the purpose of interpretation, for example, to determine what the data means in terms of to identify diseases that the subject may suffer from or be susceptible to, and/or to determine therapy or lifestyle changes that the subject may wish to adopt to eliminate and/or prevent the disease condition. For example, a subject's genetic sequence and/or its variant definition file may be analyzed to determine clinically relevant genetic markers that indicate the presence or potential of a disease state and/or the effectiveness with which a recommended therapeutic or prophylactic regimen may affect the subject. These data may then be used to provide the subject with one or more therapeutic or prophylactic regimens in order to improve the subject's quality of life, eg, treat and/or prevent a disease state.

[00161] In particular, medical science technology has evolved in tandem with advances in information technology, which have increased our ability to store and analyze medical data. Therefore, once one or more genetic variants in an individual have been identified, such variant definition file information can be used to produce medically useful information, which in turn can be used to determine, for example, using known statistical models analysis, health-related data and/or medically useful information, e.g. for diagnostic purposes, e.g., to diagnose a disease or its potential, clinical interpretation (e.g., looking for markers that represent a disease variant), whether to include subject into or excluded from various clinical trials, and for other such purposes. Because there are a finite number of disease states that are caused by genetic disorders, in tertiary processing, variants of a particular type, for example, known to be associated with the occurrence of disease states, can be refined, for example, by determining whether one or more genetic markers of the disease are included to the subject variant definition file.

[00162] Therefore, in various cases, the methods described herein may include analyzing, for example, scanning, VCF and/or generated sequence for known disease-associated sequence variants, for example, present for this reason in a database of genomic markers to detect the presence of a genetic marker in the VCF and/or generated sequence, and if present, test for the presence or possibility of a genetically determined disease state. Because there are a huge number of known genetic variations and a huge number of individuals suffering from diseases caused by such variations, in some embodiments, the methods described herein may include the formation of one or more databases linking whole genome sequenced data and/or associated with it a file identifying variants, for example, from one or multiple individuals with a disease condition, and/or searching generated databases to determine whether a particular subject has a genetic makeup that predisposes him to having such disease condition. Such a search may involve comparing one complete genome with one or more others, or a fragment of a genome, such as a fragment containing only variations, with one or more fragments of one or more other genomes, for example, in a database of reference genomes or fragments thereof.

[00163] In addition, it should be understood that the genetic sequences involved in these methods may be DNA, single-stranded DNA (ssDNA), RNA, mRNA, rRNA, tRNA, and the like. Therefore, although the present disclosure makes various references to various methods and devices for analyzing the DNA of a genome, in various examples the systems, devices and methods disclosed herein are equally suitable for performing their respective functions, such as analysis, on all types of genetic material , including DNA, ssDNA, RNA, mRNA, rRNA, tRNA and the like. Additionally, in various examples, the methods of the present invention may include analyzing a generated genetic sequence, such as DNA, ssDNA, RNA, mRNA, rRNA, tRNA, and the like, of a subject and determining from it protein variations that are most likely caused by the genetic sequence, and /or determining and/or predicting from it a possible disease state, for example caused by an error in protein expression. It should be noted that the resulting genetic sequence may represent an intron or an exon, for example, the genetic sequence may only refer to the encoded portion of the DNA, for example, where the exome is obtained, and using only known processing technologies, only the encoded regions, or unencoded regions, can be sequenced. may result in faster sequencing and/or faster processing, although subject to more complex sample preparation procedures.

[00164] Currently, such steps and analyzes described herein are typically performed in various separate or unrelated steps, often using different analytical equipment at different locations. Accordingly, in various aspects, the methods and systems of the present invention are implemented in a single device and/or at a single location, such as in conjunction with an automated sequencer or other device configured to generate genetic sequence data. In various examples, multiple devices may be used at a single location or at multiple remote locations, and in some cases, the methods may use two or more processing units deployed at two or more locations.

[00165] For example, in various aspects, a pipeline may be provided that includes performing one or more analytical functions, as described herein, on the genomic genetic sequence of one or more individuals, for example, on data obtained in a digital file format, for example, FASTQ, from an automated sequencer. A typical pipeline to be executed may include one or more sequenced genetic materials, such as part or all of a genome, of one or more subjects, wherein the genetic material may comprise DNA, ssDNA, RNA, rRNA, tRNA, and the like, and/or In some cases, the genetic material may represent encoded or non-encoded regions such as exomes, DNA episomes. The pipeline may include one or more of performing a base determination and/or error correction operation, for example, on digitized genetic data, and/or may include one or more of performing a mapping, alignment, and/or sorting function on the genetic data. In certain cases, the pipeline may include performing one or more of realignment, deduplication, base quality assessment recalibration, reduction, and/or compression and/or decompression on the digitized genetic data. In certain cases, the pipeline may include performing a variant detection operation on genetic data.

[00166] Thus, in various cases, the pipeline of the present invention may comprise one or more modules, where the modules are configured to perform one or more functions, such as a base detection and/or error correction operation, and/or a mapping, alignment, and/or sorting genetic data, such as sequenced genetic data. And in various cases, the pipeline may comprise one or more modules that are configured to perform one or more of local realignment, duplicate removal, ground truth recalibration, variant detection, reduction, and/or decompression on genetic data. Many of these modules can be executed either by software or hardware, or remotely, such as through software or hardware, such as in the cloud or on a remote server and/or bank of servers.

[00167] Additionally, many of these pipeline steps and/or modules are optional and/or can be arranged in any logical order and/or omitted entirely. For example, the software and/or hardware described herein may or may not include base detection or a sequence correction algorithm, for example where there may be concerns that such functionality may introduce statistical bias. Therefore, the system will include or not include base detection and/or sequence correction functionality, respectively, depending on the required level of accuracy and/or efficiency. And, as stated above, one or more functions of the pipeline may be used in generating the genomic sequence of a subject, for example, by reconstructing the genome from a reference. Additionally, in certain cases, the output from the pipeline is a variant definition file indicating partial or complete variants in the genome or part thereof.

[00168] Accordingly, as stated above, the output of sequencing protocols, such as one or more of the above, is typically a digital representation of a subject's genetic material, such as in the FASTQ file format. However, autorad, expressed in digital form, can also be used. In particular, the output from a sequencing protocol may contain multiple reads, where each read contains a sequence, such as a string, of nucleotides where the position of each nucleotide was determined, and a qualitative score representing the probability that the nucleotide was identified incorrectly. However, the quality of these output data can be improved through various preprocessing protocols to achieve higher quality estimates, and one or more of these protocols can be applied in the methods described herein.

[00169] For example, in some cases, the raw FASTQ file data may be processed to clean up the original base definitions obtained from the sequencer/reader, such as during a primary processing step, such as prior to the secondary processing described above herein. In particular, the sequencer/reader typically analyzes sequencing data, such as fluorescent data indicating which nucleotide is at which position, and converts the image data into a base determination with a qualitative assessment, such as where the qualitative assessment is based on the comparative brightness of the fluorescence at each position. A specialized algorithm can be applied, for example during the primary processing step, to correctly analyze these differences in fluorescence to more accurately perform base determination. As stated above, this stage may be included in a pipeline of stages and may be implemented through software and/or hardware, however, in this case it will be part of the primary processing platform.

[00170] An additional processing step may include an error correction function, which may involve attempting to account for the million to billion reads in the FASTQ file and correct some portion of the mechanical sequencing errors with information related to base determination and qualitative assessment available before any subsequent a processing step, such as a mapping, alignment and/or sorting function, etc. For example, reads in a FASTQ file can be analyzed to determine whether there are any subsequences in any reads that appear in other reads, which, due to redundant coverage, can increase confidence that the subsequences in the reads may be correct. This can be implemented by constructing a hash table containing all possible k-mers of a chosen length, k, from each read, and storing each of them its frequency, as well as which bases immediately follow it and with what probability. Then, using a hash table, each read can be continuously scanned. Since each k-mer in a particular read can be found in the hash table, an assessment can be made of whether the base immediately following that k-mer is likely to be valid or not. If the probability of this is low, then it can be replaced by the one that most likely follows it, based on the table. Successive k-mers for that read will then contain the correct base as the value at that position and the process will repeat. This can be highly effective at correcting errors, as oversampling ensures that accurate statistics are collected to predict what will follow each k-mer. However, as noted above, such adjustments may introduce statistical bias into the system, for example due to erroneous adjustments to the data, so these procedures can be skipped if necessary.

[00171] Accordingly, according to aspects of the present invention, in various examples, the methods, devices and/or systems of the present invention may involve obtaining read data that either has or has not been pre-processed, for example due to being obtained directly from an automated sequencer FASTQ file, and subjecting the resulting data to one or more of mapping, alignment, and/or sorting functions. Performing such functions can be useful, for example, as indicated above, in various examples, due to the fact that the generated data is usually generated by various automated sequencers, for example, reads, are significantly less in length than the entire genomic sequence being analyzed, and since the human genome is usually has many repeat regions and is known to have different repeat patterns, there may be many places at which any given sequence of reads could correspond to a segment of the human genome. Therefore, given all the possibilities a given read may match a genome sequence, for example due to various repeated sequences in the genome, etc., the raw read data may not explicitly indicate which of the possibilities is actually the correct position from which it was derived . Thus, for each read, it will be necessary to determine where in the genome the read actually maps. In addition, it may also be useful to determine the sequential alignment of the reads to determine the actual sequence identity of the subject, and/or it may also be useful to determine the chromosomal position for each portion of the sequence.

[00172] In various examples, the methods of the present invention may be directed to mapping, aligning and/or sorting the raw read data of a FASTQ file to find all locations where the read is likely to be aligned and/or determining the actual sequence identity of a subject, and /or determination of the position in the chromosome for each part of the sequence. For example, mapping can be used to map generated reads to a reference genome and thus find the position at which each read appears to have a good match to the genome, e.g., finding all locations that might have a good score for the alignment any given read relative to the reference genome. Thus, mapping may involve taking one or more, for example, all, of the raw or preprocessed reads from a FASTQ file and comparing said reads to one or more reference genomes, and determining where said read might match the reference genome(s) . In its basic form, mapping involves searching for the position(s) in a reference genome where one or more of the FASTQ reads obtained from the sequencer matches.

[00173] Likewise, alignment can be used to evaluate possible positions of an individual's reads relative to a reference genome window to determine where and how the sequence of reads best aligns to the genome. However, performing an alignment can be difficult due to substitutions, insertions, deletions, structural variants, and the like, which can prevent the read from aligning accurately. Thus, there are several different ways to obtain an alignment, but this may require changes to the read, with each change that needs to be made to obtain a proper alignment resulting in a lower confidence score. For example, any given read may have substitutions, insertions, and/or deletions when compared to a reference genome, and these variations must be taken into account when performing the alignment.

[00174] Accordingly, along with the predicted alignment, an estimate of the probability that the predicted alignment is correct can also be given. This score indicates the best alignment for any given read among the many positions at which the read could be aligned. For example, an alignment assessment is based on how well a given read matches a potential mapping position, and may involve stretching, densifying, and changing parts of the read to produce the best alignment.

[00175] The score will reflect all changes to the read to match the reference. For example, creating an alignment between a read and a reference may require the insertion of one or more gaps into the read, where the insertion of each gap represents a deletion in the read relative to the reference. Likewise, it may be necessary to perform a deletion in a read, with each deletion representing an insertion in the read relative to the reference. In addition, various bases may need to change, for example due to one or more substitutions. Each of these changes is made to ensure that the read(s) are more accurately aligned with the reference, but each change is made at the cost of quality scores, which are a measurement of how well the entire read matches some region of the reference. The reliability of such qualitative estimates is then determined by observing all positions to which a read can be mapped and comparing the scores at each position, and selecting the location with the highest score. In particular, when there are many items with a high quality score, then confidence is low, but when the difference between the first and second best scores is large, then confidence is high. At the end, all proposed reads and confidence scores are evaluated and the best match is selected.

[00176] Once the reads are assigned positions relative to the reference genome, which consists of determining which chromosome the read belongs to and its offset from the beginning of that chromosome, they can be sorted, for example, by position. This allows subsequent analyzes to take advantage of the various oversampling protocols described herein. All reads that overlap a given position in the genome may be adjacent to each other after sorting and can be organized into a cluster and easily examined to determine whether most of them agree with the reference value or not. If not, as stated above, the option can be checked.

[00177] As stated above, a FASTQ file obtained from a sequencer consists of many, for example millions to a billion or more, reads consisting of short strings of nucleotide sequence data representing part or all of an individual's genome. Mapping typically involves mapping reads to all locations in the reference genome where there is a match. For example, depending on the size of the read, there may be one or many locations where the read substantially matches the corresponding sequence in the reference genome. Accordingly, the mapping and/or other functions described herein can be performed to determine which of all possible locations where one or more reads may match a reference genome is actually the true location to which they are mapped.

[00178] Each read can be compared to each position in the 3.2 billion reference genome to determine where the reads, if matched, match the reference genome. This may be accomplished, for example, when the read length reaches about 100,000 nucleotides, about 200,000 nucleotides, about 400,000 nucleotides, about 500,000 nucleotides, and even about 1,000,000 or more nucleotides. However, when the reads are significantly shorter in length, such as when there are 50 million reads or more, such as 1 billion reads, this process can take a very long time and require a large amount of computing resources. Accordingly, there are several methods, such as those described herein, that have been developed to align FASTQ reads to a reference genome in a much faster manner. For example, as described above, one or more algorithms may be used to map one or more reads generated by a sequencer, such as in a FASTQ file, and correlate them with a reference genome to determine where in the reference genome the read in question could potentially be mapped.

[00179] For example, in various methods, a reference index is generated so that reads or parts of reads in the index can be searched, retrieving location indicators in the reference to map the reads to the reference. Such a reference index can be constructed in various forms and accessed in various ways. In some methods, the index may contain a tree of prefixes and/or suffixes. In various other methods, the index may include a Burrows-Wheeler reference transformation. In other methods, the index may contain one or more hash tables, and the hash function may be performed on one or more portions of the reads for the purpose of mapping the reads to a reference. In various examples, one or more of these algorithms may be performed sequentially or simultaneously to accurately determine where one or more, for example, a significant portion of the reads or each read, of the reads correctly matches the reference genome.

[00180] Each of these algorithms may have advantages and/or disadvantages. For example, a suffix and/or prefix tree and/or a Burrows-Wheeler transform can be applied to sequence data such that a reference genome index is constructed and/or queried as a tree data structure, where starting with a single-base or short subsequence of the read, the subsequence is gradually expanded within read, accessing the index at each step, tracing the path in the tree data structure until the subsequence is sufficiently unique, for example, the optimal path length is reached, and/or the leaf node of the tree structure is reached, where the leaf or last tree node reached shows one or more positions in the reference genome from which the read begins. Thus, these algorithms typically do not have a fixed length for sequence reads that can be mapped by index searching. A hash function often uses a fixed-length unit of comparison, which can be the length of the entire read, but is often several times the length of some part of the read, called the seed. Such seeds can be shorter or longer, but unlike prefix and/or suffix trees and/or Burrows-Wheeler transformations, the read seeds involved in the hash function have a pre-selected, fixed length.

[00181] A prefix and/or suffix tree is a data structure that is constructed from a reference genome such that each edge (link) connecting a parent node to a child node is labeled or associated with a nucleotide or nucleotide sequence, and each path from a root element through various edges and nodes corresponds to some subsequence of the reference genome. A node reached in this way is implicitly associated with a reference sequence corresponding to the path from the root element. Starting from the root element, the prefix tree corresponds to the forward continuation of the reference genome sequence, while the subsequences of the suffix tree correspond to the backward continuation of the reference genome. Both kinds of trees, prefix and suffix, can be used in hybrid prefix/suffix algorithms, so that subsequences can line up in both directions. Prefix and suffix trees may also contain additional edges, such as transitions from a node associated with one reference subsequence to another node associated with a shorter subsequence.

[00182] For example, queries against a tree data structure serving as an index to a reference genome can be performed by identifying a path in the tree corresponding to a subsequence of the read to be mapped. The path is built as nucleotides are added to the sequence, each of which determines the next node as it traverses the tree, until a unique sequence is obtained. This unique sequence may also be called a seed, which may represent a branch and/or root of a tree structure of the sequence. Alternatively, descent through the tree structure can be stopped before the aggregate sequence is completely unique, so that the resulting primer can correspond to multiple locations in the reference genome. In particular, a tree can be constructed for each starting position of a reference genome, in which case the resulting reads can be compared to the branches and/or roots of the tree, so that the tree can be traversed according to these sequences to determine where in the reference genome the read corresponds. More specifically, reads from a FASTQ file can be compared to branches and roots of a reference tree, allowing, if a match, their position in the reference genome to be immediately determined. For example, a sample read can be used to traverse the tree until the sequence retrieved is unique enough to determine that the read actually aligns with a particular position in the reference genome. In particular, one can go down the tree to a leaf node.

[00183] A disadvantage, however, of such a prefix/suffix tree is that the huge data structure must be accessed many times as the tree is traversed in order to map reads to a reference genome. On the other hand, the advantage of a hash table, as detailed later in this application, is that once the table is obtained, one call to it is usually sufficient to determine where, and whether, there may be a match between read and reference genome. In general, it will take many, such as 5, 10, 15, 20, 25, 50, 100, 1,000, or more, etc., calls to the prefix/suffix tree to determine if there is a match. Further, due to the double helical structure of DNA, it may also be necessary to construct and examine a reverse complement tree, as the reverse complement to a reference genome may need to be found. Considering the above, a data tree can be described as a structure built on the basis of a reference genome, which is then compared with reads sequenced from the subject's DNA, however, it should be understood that the data tree can be initially from both the reference sequence and the reads of the sample, or both and the other, after which a comparison is made of one with the other, as described above.

[00184] Alternatively, or in addition to applying a tree of prefixes or suffixes, a Burrows-Wheeler transform can be applied to the data. For example, the Burrows-Wheeler transform can be used to store a tree data structure, abstractly equivalent to a prefix/suffix tree, in a compact format, such as the space allocated for storing a reference genome. In various embodiments, the data stored is not a tree structure, but rather the reference sequence data is stored in a linear list, which can be arranged in different orders in order to transform it in a very specific way, so that the accompanying algorithm allows you to search for regions in the sequence that match reads the sample, thus effectively traversing the tree. The advantage of the Burrows-Wheeler transform, compared to a prefix tree or a suffix tree, is that it typically requires significantly less memory to store data, and compared to hash tables is that it supports variable seed length, and therefore, a unique sequence can be identified in it and a match can be found. However, just as in the case of a prefix/suffix tree, how many nucleotides of a given sequence were required to determine uniqueness, or to map to a relatively small number of reference positions, determines the length of the primer. At the same time, for a hash table, all seeds have the same, predefined length. However, the disadvantage of the Burrows-Wheeler transform is that it generally requires accessing the data structure several times, for example two or more times, just as when descending a tree sequentially.

[00185] Alternatively, or in addition to applying one or both of the prefix/suffix tree and/or Burrows-Wheeler transformation methods to the reference genome and sequence data of a subject to learn where one maps to the other, another such method involves creating a hash table index and/or computing a hash function. The hash table index may be a large reference structure that is constructed from the reference genome sequences, which may be compared to one or more parts of the read to determine a possible match between one and the other. Similarly, a hash table index can be built on parts of a read, which can then be compared with one or more reference genome sequences, which in turn allows a possible match to be determined.

[00186] More specifically, in any of the mapping algorithms described herein suitable for implementing steps of the methods described herein, one or all three mapping algorithms, or others known in the art, may be used in software or hardware to map one or more sequences of a sequenced DNA sample to one or more reference genomes. As described in great detail below, all of these actions can be carried out either in software or in hardware, for example using integrated circuits such as a chip, perhaps as part of a printed circuit board. For example, the execution of one or more of these algorithms may be built into a chip, such as an FPGA (Field Programmable Gate Array), an ASIC (Application-Specific Integrated Circuit), or a Structured ASIC (Application-Specific Integrated Circuit), and can be optimized for better performance through implementation.

[00187] Additionally, one or more, e.g., two or all three of these mapping functions may form a module, such as a mapping module, which may form part of a system, e.g., a pipeline, that is used in a process to determine the complete actual sequence of a genome, or its part, subject. The output obtained after applying the mapping function may be a list of probabilities that one or more, for example, each read maps to one or more reference genomes. For example, the output for each mapped read can be a list of possible reference genome locations to which the mapped read corresponds. In various embodiments, it is possible to search for an exact match to a reference genome, if not the entire read, then at least a portion, such as a primer. Accordingly, in various embodiments, it is not necessary to have an exact match of all parts of all reads to all parts of the reference genome.

[00188] Further, one or all of these functions may be programmed to provide an exact and/or approximate match and/or the ability to edit the results. Therefore, all of these processes can also be configured to search for partial matches if desired, for example according to a pre-selected possible discrepancy, such as an 80%, 85%, 95%, 99%, or greater match. However, as detailed below, searching for a partial match can be significantly more time- and computationally expensive because it may require many edits, such as a single nucleotide polymorphism (SNP) or the insertion or deletion of one or more nucleotide bases, for example, it may 1 or 2 or 3 or 5 or more edits are required until an acceptable fit is achieved. Such edits are likely to be widely used in the implementation of hashing protocols, or the construction of prefix and/or suffix trees and/or the Burrows-Wheeler transform.

[00189] As for hash tables, they can be obtained in different ways. In one embodiment, a hash table is constructed by breaking the reference genome into segments of standard length, such as primers of about 16 to about 30 nucleotides or more in length, or such as about 18 to about 28 nucleotides, reducing them to a table format that can be searched , after which an index is compiled of all reference segments from which the DNA is sequenced, for example, 1 or more reads, or part thereof, can be compared to determine the match. More specifically, a hash table index can be generated by breaking the reference genome into segments of nucleotide sequences of known, equal length, such as a primer, and storing them in random order in separate table cells. This can be done for part or the entire reference genome to construct an actual reference index, which can be used to compare parts of the reference genome with parts of one or more reads, for example from a FASTQ file, to determine a match.

[00190] This method can then be repeated in approximately the same manner for a portion, eg, a majority of the reads in the FASTQ file, to generate seeds of a suitable, eg, selected length. For example, reads from a FASTQ file can be used to create seeds of a predetermined length, which can be converted to binary form and fed into a hash function, and queried in the hash table index, where the binary form of the seed can be matched with the binary representation of the reference segment genome, which gives an idea of the location of the genome where there is a match with the sample seeds.

[00191] For example, if a read is approximately 100 nucleotide bases in length, a typical primer may be about two or three times smaller, such as about 27 to about 30 nucleotide bases in length. Therefore, in such an implementation, several, approximately 3 or 4, seeds are required, depending on the read length and/or seed lengths, to completely cover the read. Each primer can be converted to binary form and/or then fed into a hash table, after which a result can be obtained indicating the corresponding location in the reference genome. In such implementations, there is no need to compare the entire read with each possible position in the reference genome; it will be sufficient to compare parts of the reads, for example one or more seeds for each read, with an index containing portions of the reference genome that are equivalent in size. Therefore, in various embodiments, the hash table can be formed in such a way that one memory access is sufficient to determine the location in the reference genome with which the primer has a match, and, therefore, the sample read. However, in some embodiments, it may be necessary to apply the hash function and lookup to one or more overlapping seed sections from the same read. In these embodiments, the primers can be selected such that at least a portion of their sequences overlap. This can help avoid machine and/or human errors, or differences in the subject's genome and the reference genome, and facilitate finding an exact match.

[00192] In some embodiments, the construction of the hash table, as well as one or more of the various comparisons, are performed through a hash function. The hash function is partly a data scrambler. It takes data as input and outputs what at first glance looks like a random sequence. In this implementation, a hash scrambler breaks the reference genome into preselected segments and randomly places them in a hash table. Alternatively, the data storage space may be segmented and/or different weights may be assigned for storage purposes. More specifically, a hash function is a function that takes any input and outputs a number, such as the output of binary patterns, where the output values can be anything except that given the same input value, the output value will also always be the same. Therefore, if two sequences fed into a hash function are essentially the same, because they are not exactly the same, the output will be two completely different values.

[00193] Further, since genetic material can be composed of four nucleotide bases, for example, “A”, “C”, “G”, and “T” (or “U” in the case of RNA), individual nucleotide sequences, such as reference segments and/or reads, or parts thereof, that will be input to the hash function can be digitized and represented in a binary format, for example one in which each of the 4 nucleotide bases is represented as a two-bit digital code, for example “A” = 00, “C” = 01, “G” = 11, and “T”/”U” = 10. In some embodiments, it is this binary “seed” value that is randomly placed in the hash table at a known location that has the value equal to its binary representation. The hash function thus serves to partition the reference genome into binary representations of the reference primers and inserts each binary primer into a random location, such as a hash table cell, based on a numerical value. Along with a digital binary code, such as an access code, each cell may contain the actual entry points where the segment was taken from the actual reference genome, such as a reference position. The reference position may thus be a number indicating the position of the original reference primer in the genome. The same can be done for overlapping positions that are placed randomly in the table but in known locations, such as through a hash function. In a method such as this, a hash table index may be obtained, the index including a digital binary code for part or all of a variety of segments of one or more reference genomes that may be referenced by one or more sequences of genetic material, for example, one or more reads, or parts thereof, from one or more entities.

[00194] When the hash table and/or function is implemented as a module, such as one of the pipeline modules, in software (so that the bit width is equal to 2x the number of nucleotide bases in the seed as described above) and/or hardware, as referenced above , the hash table can be constructed so that the binary representation of the reference read seeds has any desired width in bits. Since primers can be long or short, binary representations can be larger or smaller, but generally the length of the primer should be chosen so that it is long enough to ensure uniqueness, but not too long where it becomes too difficult to find matches between reference genome primers and sample read seeds due to the presence of errors or a large number of variants. For example, as stated above, the human genome consists of up to approximately 3.1 billion pairs, and a typical read can be on the order of 100 nucleotide bases in length. Therefore, the length of an effective primer can be from about 16 or about 18 nucleotide bases or less to about 28 or 30 nucleotide bases or more. For example, in some embodiments, the read length may be a segment of 20 nucleotide bases in length. In other embodiments, the read length may be a segment of 28 nucleotide bases in length.

[00195] Therefore, if the length of the primer is a segment of 20 nucleotide bases, then in the digital representation the data will be 40 bits long, for example, a 40-bit digital representation of the primer. For example, if two bits are selected to represent a nucleotide, such that A = 00, C = 01, G = 10, and T = 11, a primer 20 nucleotide bases long x 2 bits per nucleotide base = 40-bit (5-byte ) vector, for example, a number. If the length of the primer is 28 nucleotide bases in length, the digital, for example, binary representation will be a 56-bit vector. Therefore, if the length of the primer is approximately 28 nucleotide bases, it will take 56 bits to obtain a digital representation of the 28-base primer. More specifically, if the binary representations of the reference genome seeds are randomly distributed across the hash table and are 56 bits long, another 56 bits can be used to represent the read seeds that are to be compared with the reference genome seeds. These 56 bits can be fed into a polynomial transform, which produces 56 bits with a 1:1 mapping. Without increasing or decreasing the number of bits in the output, this transformation randomizes the storage location of adjacent input values, so that different seeds are evenly distributed across all possible storage locations. This also helps minimize the number of collisions between values that the hash function ends up in the same storage locations. In particular, in a typical hash table implementation described in this application, only a portion of 56 bits are used to access the table, with the remaining bits stored for further confirmation of the match. If a hash function were not used, a huge number of patterns would have the same address bits and different storage bits, and would have to be stored in the same places in the table.

[00196] More specifically, there are similarities between the way the table is populated, for example, by software and/or hardware seeding a reference genome being randomly placed into a hash table, and the way the table is accessed when hashing seed reads, namely , access to the hash table in these cases occurs in the same way. Consequently, reference genome seeds and read seeds that have the same binary code will end up in the same place in the table, for example, a memory area with the same address, since it is accessed in the same way, for example, for the same input vectors (pattern). This is the fastest way to find matches in patterns. Each table access takes approximately constant time. This can be contrasted with the Burrows-Wheeler method, which requires multiple checks (the number may vary depending on how many bits are required to find a unique sequence) per query to find a single match, or the binary search method, which requires log ₂ (N ) checks, where N is the number of seeds in the table.

[00197] Further, although a hash function can break the reference genome into segments of seeds of any length, for example, 28 nucleotide base pairs, and can then translate these seeds into a digital, for example, binary 56-bit representation, it does not necessarily need to have access to all 56 bits at the same time in the same way. For example, the hash function may be implemented such that the address of each seed is determined by a number less than 56 bits, such as numbers from about 20 to about 45 bits, from about 25 bits to about 40 bits, from about 28 bits to about 35 bits, including approximately 28 bits to approximately 30 bits can be either the original key or the hash table access address.

[00198] For example, in some embodiments, from about 26 to about 29 bits can be used as the primary access key to a hash table, leaving out the remaining 27 to 30 bits, which can be used as a means of double checking the first key, for example, if both the first and the second keys end up in the same cell of the hash table, then it is relatively clear that they belong to the specified location. Thus, to save space and reduce memory consumption and/or processing time of the hash module, for example, when the hash table and/or hash function is implemented in hardware, from about 26 to about 29 bits of the original bits may be retained to represent the primary access key. 56 representing the digitized seed of a specific sequenced read, and the remaining 27 to 30 can be used for double control.

[00199] More specifically, in various embodiments, from about 26 to about 29 bits of the 56 bits of the binary representation of the reference seed can be used to obtain the primary address, which is defined 26 to 29 and may have a randomized location in the hash table, which in turn can be filled with the location to which the reference seed originally belonged, as well as the remaining 27 to 30 bits of the seed, so that an exact match can be confirmed. Query seeds representing the subject's genome reads converted to binary form can be hashed with the same function so that they are also represented by 29 bits containing the primary access key. If the 29 bits representing the reference seed are exactly the same as the 29 bits representing the request seed, they will both be directed to the same location in the hash table. If there is a complete match with the reference seed, it is expected that a record containing the same remaining 27 to 30 bits will be found in that location. In this embodiment, the 29 reserved address bits of the reference sequence are used to reference the table to determine the position in the reference sequence to which the query read from which the query seed was extracted is aligned.

[00200] However, for the remaining 27 to 30 bits, these bits can represent a secondary access key, which can also be imported into a hash table to validate matches found against the 26 to 29 bit primary key. Since a hash table is a perfect 1:1 conversion of a 28 nucleotide base/56 bit sequence, and only 26-29 of these bits are used to determine the address, these 26-29 bits were first checked and thus determined the correct address on the first pass. These data therefore do not need to be confirmed. However, the remaining approximately 27 to approximately 30 bits of the secondary key need to be verified. Accordingly, the remaining approximately 27-30 bits of the query seed are inserted into the table as a means of completing the match search. This implementation can be shorter than sorting full 56-bit keys, and thus saves space and reduces the amount of memory required and module processing time.

[00201] Thus, a hash table can be configured as an index where the sequences of one or more reference genomes, divided into sequences of a predefined length, for example 28 nucleotide bases, are randomly assembled into a table, and the sequence of one or more reads or their “seeds” portions" obtained by sequencing a subject's DNA or RNA can be run through a hash table index, for example according to a hash function, to determine the position of the seeds in the table, allowing one or more positions, for example a location in a reference genome, to be obtained when the sample is primed corresponds to the location of the reference genome. Using brute force search to scan the reference genome in order to find locations that match the primers would require searching through 3 million positions. However, with hashing, determining where each seed matches matches takes approximately constant time. Often, the location of the match can be confirmed in one call to the table. In the case where several seeds fit one place in the table, several additional calls can be made to clarify the result. Therefore, although a 100-bp read could potentially correspond to 30M positions in the reference genome, a hash table and a hash function can quickly determine whether a read is included in the reference genome. Accordingly, when using an index, hash tables are not needed in determining the position to which the read is aligned.

[00202] As stated above, chromosomes have a double helix structure, which consists of two opposing complementary strands of nucleic acid sequences that are linked to each other to form a double helix. For example, when forming a double helix structure, these complementary base pairs bind to each other according to the following formula: “A” binds to “T” and “G” binds to “C.” Accordingly, this produces two equal and opposite chains of nucleotide sequences that are complementary to each other. More specifically, the bases of a nucleotide sequence on one strand are mirrored by their complementary bases on the other strand, resulting in two complementary strands. However, DNA transcription occurs in only one direction, starting at one end of the DNA and moving to the other. Accordingly, it turns out that for one DNA strand, transcription occurs in one direction, and for its complementary strand, transcription occurs in a different direction. Accordingly, it turns out that two strands of DNA sequences are inversely complementary to each other, which means that if the sequence order of one DNA strand is compared with another, then we will see two strands in which the letters denoting nucleotides in one strand are replaced by their complementary ones in the other, for example , the letters “A” to “T”, and “G” to “C” and vice versa, and their order is reversed.

[00203] The double helical structure of DNA means that at the preparatory stage before sequencing the DNA is separated, i.e. denatured, separated into two separate strands, and then lysed into smaller fragments of a predetermined length, for example, 100-300 bases in length, which are then sequenced. It is possible to separate the strands before sequencing so that only one strand is sequenced, but typically the DNA strands are not separated and both DNA strands are sequenced. Accordingly, in this case, approximately half of the reads in the FASTQ file will be reverse complementary.

[00204] Of course, both strands of the reference genome, for example, complement and reverse complement, could be processed and hashed as described above, however, this would double the size of the hash table, and would take twice the time to run the hashing function, e.g. this could require twice as much computation to compare both the complement and reverse complement sequences of two genomic sequences. Accordingly, to save memory space, reduce the need for computing resources and/or reduce processing time in various options, only one strand of model DNA needs to be stored as a reference in the hash table.

[00205] However, because under conventional sequencing protocols, when the two strands of a subject's DNA are not separated from each other, any read generated from the sequenced DNA may be either complement or reverse complement, respectively, it may be difficult to determine which strand is processed: complement or reverse complement. More specifically, in various embodiments, since only one strand of the reference genome needs to be used to generate a hash table, half of the reads generated by the sequencing protocol may not match that particular strand, e.g., complementary or inversely complementary, of the reference genome model, since half the time is being processed The read is reverse complementary to the hashed segments of the reference genome. Accordingly, only reads generated from one DNA strand will have a match with the indexed sequences of the reference genome, and reads generated from the other strand, theoretically, will be their reverse complements and will not have matches to any regions of the reference genome. Further, an additional complication may be that for any given read that is the reverse complement of a conserved reference genome, that read may still produce an erroneous match to part of the reference genome, for example, by pure chance. In view of the above, for effective mapping in various variants, it is necessary not only to determine which place in the reference genome the read matches, but also to determine whether the read is reverse complementary. Accordingly, the hash table and/or function module should be designed to minimize these difficulties and/or the types of errors resulting from these difficulties.

[00206] For example, as noted above, in one embodiment, both the complement and the reverse complement for a reference genome can be placed in the hash table, such that for each read or its reverse complement of a subject's sequenced DNA, a match can be made on the corresponding strand of the reference genomic DNA . In this case, for any given primer in a read, the primer should theoretically have a match with one of the strands, complement or reverse complement of the reference, errors and deviations not taken into account. However, storing both strands of the reference genome in a hash index may require twice as much storage space (for example, it may require 64 gigabytes instead of 32 gigabytes), and may also require twice as many processing resources. Further, this solution does not solve the problem of palindromes that can have matches in both directions, for example, complementary and reverse complementary strands.

[00207] Accordingly, while a hash table index may be constructed to include both strands of a reference genomic sequence, in various embodiments, a hash table may be constructed to include only one strand of a model genome as a reference. This can be useful because storing a hash table in memory will require half the storage and/or processing resources of what would be required to store and process both threads, and therefore the lookup will also take less time. However, storing only one strand of the genome as a reference can be problematic because, as stated above, when a subject's DNA being sequenced is double-stranded, it is usually not known from which specific strand each particular read was generated. Accordingly, in this embodiment, the hash table must be generated taking into account the fact that the mapped read can be from any strand and, accordingly, can be the complement or reverse complement of the conserved reference genome segments.

[00208] Accordingly, in various embodiments, such as when only one orientation of a seed from a reference is placed in a hash table, when executing a hashing function with seeds generated from reads from a FASTQ file, one can first search for the seed in its current orientation and/or can then be converted to reverse complement and searched with reverse complement. This may require two hash index searches, i.e. twice as large, but either the primer or its inverse component must match the complementary segment in the reference genome, provided there are no errors or variations, and this should reduce the overall need for computing resources, for example, due to less memory usage, as well as reduction time, for example, when not such a large number of sequences are compared.

[00209] More specifically, as in the case where the primer in one particular orientation consists of 28 nucleotides, which for example are represented digitally in 56 bits of binary format as described above, the primer can be converted to the reverse complement, and the reverse complement as well can be represented digitally in 56-bit binary format. The binary format for each representation of a primer sequence and its complement produces a number, such as an integer, having the value represented by that number. These two values, for example two integers (numbers), can be compared and the number with a higher or lower value, for example a higher or lower absolute value, can be chosen as the canonical orientation option and it will be stored in the hash. table and/or processed by a hash function. For example, in some embodiments, the number with the higher value may be selected for processing by the hash function.

[00210] Another method can be used to design primers, in which each primer consists of an odd number of bases. As a canonical orientation, you can choose threads in which the base in the middle is “A” or “G”, but not “T” or “C”, or vice versa. The hash function can then be applied to seeds that meet the canonical orientation requirements. Thus, to compare to determine which has a higher value, only the two bits representing the middle base need to be compared, and only those 2 bits of that sequence can be used for searching. Accordingly, only the bits representing the two middle bases need to be taken into account. This usually works well because the primer is of odd length and therefore its middle base is always converted to the reverse complement. However, while this may work for odd-length seeds, hashing higher or lower value seeds as described above should work for seeds of any length, even though doing so may require processing, such as searching for more bits of data .

[00211] These methods can be performed on any number of seeds, for example, all reference seeds, and/or any number of seeds, for example, all obtained from all or a portion of the reads from the FASTQ file. About half the time, the binary representation of primers of a given orientation, such as complement, will have a higher value, and approximately half the time, the binary representation of primers of the opposite orientation, such as reverse complement, will have a higher value. But when considering binary numbers, the hash table will always contain those whose value is greater. For example, you can compare binary integers for each read and its complement, and the first 1 encountered is selected as a thread to be stored in a hash table and/processed by the hashing function. If both threads contain the first 1 in the same position, then the thread with the second next 1 is selected, etc. Of course, you can also select a read with a lower value, in which case you select the thread with the first 0 in order or the first 0 with a higher number in order. An indicator, such as a flag, can also be included in a hash table, where the flag indicates which orientation, complement or reverse complement, is stored, representing the stored and/or hashed strand, such as the 1RC flag, for the reverse complement.

[00212] More specifically, when implementing a hash function and evaluating a hash table, the primers from the genomic reference DNA and the primers from the sequence data reads are subjected to the same operations, for example, converted to binary form and compared with their reverse complement, when In this case, integers with higher or lower values are chosen as canonical orientations, a hash function is applied to them, and they are passed to a hash table to be looked up and matched against each other. However, since the same operation is performed in essentially the same way on reference and read sequences, the same records will be obtained if two sequences, the reference seed and the subject seed, start with the same sequence, even if one of them is reverse complement, they will be sent to the same cell in the hash table.

[00213] Accordingly, if some primer in a reference sequence having that sequence in a particular orientation is binned and hashed, and then a primer derived from a sample read having the same sequence but in the reverse orientation, for example, reverse complement, and it is subjected to the protocols described above, by virtue of the methods disclosed above, when a binary value is determined and a hashing function is performed, the lookup will be directed to the same address in the hash table as if the hash function was performed first with a complementary seed. Accordingly, with this approach, it does not matter what orientation the processed seed has, since it will always be directed to the same address.

[00214] Accordingly, with this approach, the disclosed methods can hash and therefore determine the position of a primer in a table regardless of its orientation, and due to a flag in the entry, it will also be known if any primer is reverse complement. For example, it will also be known if the seed is flipped relative to the original orientation in the reference, and it will also be known if the prime is flipped relative to the original orientation in the subject's read. Accordingly, if the same decision was made in both cases, the read and reference have the same orientation. However, if one side is flipped and the other is not, we can conclude that the read is captured in a reverse complementary orientation relative to the reference. Accordingly, using a hash table, it is possible to determine to which place in the genome a given read or part of it, for example, a primer, corresponds and whether it is reverse complementary. Further, it should be understood that while the above applies to generating a hash table from a reference genome and performing various ancillary hash function processes with seeds generated from reads, such as a FASTQ file, the system can also be structured such that the hash index tables are generated from primers derived from the subject's sequenced DNA reads, and the various hash function auxiliary processes described herein are performed on primers generated from the reference genome.

[00215] As stated above, the advantage of using a hash table and/or a hash function is that when using primers, the majority of the sequenced DNA reads can be mapped to the reference genome, often in a single hash table search, and in a variety of variations , not all seeds obtained from a read need to be hashed and/or searched. The primers can be of any suitable length, e.g., can be relatively short, e.g., 16 nucleotides or less in length, e.g., about 20 nucleotides, e.g., about 24 nucleotides, e.g., about 28 nucleotides, e.g., about 30 or about 40, or about 50, or 75, or about 100 nucleotides, or even up to 250, or 500, or 750, or even 999, or even about 1000 nucleotides in length; or may be relatively long, such as greater than about 1000 nucleotides, or greater than about 10,000, or greater than about 100,000, or greater than 1,000,000, or greater than nucleotides in length. However, as described above, there are some disadvantages to using seeds, such as those associated with the hash table, particularly with regard to selecting reads of appropriate length.

[00216] For example, any suitable length of seed can be used in a mapping function, but there are advantages and disadvantages to using relatively long and relatively short length seeds. For example, the shorter the seed length, the less likely it is to include an error or variant that could prevent a match from being found in the hash table. However, the shorter the sieve length, the less unique it is, and the greater the likelihood of a match between the reference genome primer and the primers derived from the subject's sequenced DNA reads. Further, the shorter the seed, the more searches need to be carried out by the hash function, which takes more time and requires increased computing power.

[00217] On the other hand, the longer the seed, the higher its uniqueness, and the less likely it is that there will be multiple matching positions between the reference and query seeds. Further, the longer the seed, the fewer seeds are needed for each read, the fewer searches and, accordingly, less time and fewer calculations are required. However, the longer the primer, the more likely it is that the primers derived from sequenced DNA contain an error, such as a sequencing error, and/or may include sequence deviations that prevent a match from being determined. Additionally, longer reads may have the additional disadvantage of being more likely to fall at the end of the read and/or the end of the chromosome. Accordingly, if the primer length is only 20 - 100 nucleotides, there may be a few matches in the hash table, but if the primer length is 1000 nucleotides or more, there will be significantly fewer matches, but there may be no matches at all.

[00218] There are a number of methods to minimize these disadvantages. One way is to ensure that sufficient oversampling is generated during DNA processing steps prior to sequencing. For example, if it is known that there is typically at least one variation (variant) for every 1000 base pairs, the length of the primer can be chosen to maximize matches while minimizing mismatches due to possible errors and/or variations. Additionally, the use of oversampling, for example during the pre-sequencing and/or sequencing steps, can be used as an additional method to minimize various problems associated with the use of primers, for example, in a hash function.

[00219] As stated above, redundant sampling creates aliasing. Strands are groups of reads that map overlappingly to broadly the same location in the genome. For most reads from a sample, such layering may not be necessary, as is the case when the reads, and/or the seeds generated from them, do not include a variant and/or do not map to multiple positions in the hash table (e.g., are not completely duplicated in the genome ). However, for those reads and/or seeds that may include a variant and/or error and/or other mismatch between the seed and/or read and the reference genome, layering for each given genomic region may be beneficial. For example, even though only exact hits between a primer generated from a sample genome read are needed to map the sample read to a non-reference genome, there may still be a technical error or a true variant in the sample DNA sequence that may prevent determination exact match between read and reference, which often makes the formation of overlapping layers in the pre-sequencing or sequencing phase useful.

[00220] For example, for those cases in which the sample seeds actually contain a variant or error, stacking reads can be useful to distinguish actual deviation from technical and/or chemical errors. In this embodiment, layering can be used to determine whether an apparent deviation is a real deviation. For example, if 95% of the reads in a stack contain a “C” at a certain position, then the rest determine the correct solution, even if the reference genome contains a “T” at that position. In this case, the mismatch may be due to a SNP, for example, a replacement of “T” with “C” at this genomic position, in cases where the subject's genetic code does differ from the reference. In this embodiment, layering depth can be used to compare overlapping portions of a layer at a deviation position, and based on the proportion of reads in the layering that have that deviation, it can be determined whether the deviation is actually due to an actual deviation in the sample sequence. Accordingly, the actual sequence of reads that best fits the genomic sequence can be determined in part based on the information reflected by the layering depth. However, the use of layering has the disadvantage that it requires more time to process all the redundant reads and/or the seeds they generate.

[00221] Another way to minimize the problems inherent in short or long reads is to use a secondary hash table in parallel or in conjunction with the first, for example, a primary hash table. For example, the second hash table and/or hash function can be used for seeds that do not have a single hit in the primary hash table, or for seeds with multiple hits in the primary hash table. For example, comparing one seed to another may produce several possible results. In one case, there may be a lack of hits, for example, a lack of matches between two sequences, indicating a possible error or deviation, for example, of a read seed from a subject compared to a seed derived from a reference genome. Either one or many matches may be found. However, finding a large number of matches can create a problem.

[00222] For example, with respect to a primary hash table, if each seed in the hashed reference sequence appears only a small number of times, such as once, twice or three times, etc., then there may be no need for a secondary hash table and /or hashing functions. However, if one or more seeds appear a greater number of times, for example 5, 10, 15, 20, 25, 50, 100, 1,000 or more times, this may be problematic. For example, there are known regions of the human genome that have been determined with mathematical significance to be repeated many times. Accordingly, mapping any primer to one of these positions can essentially automatically map to a plurality of these positions, as is the case when the primer contains nucleotides of overlapping sequences. In this case, it can be difficult to determine which of all possible variants this primer aligns with. However, as these repeat regions are known and/or when they become known, any seed, any seed typically mapping to one or more of these regions, can be extracted and allocated to a secondary hash table for processing by the first or secondary hashing function, which respectively , allows you to avoid wasting time and computing power trying to use the primary hashing function to determine something that with a high probability cannot be determined.

[00223] More specifically, comparisons of reference genomic sequence primers with primers generated from the subject's genome reads can yield any number of match positions from 1 to hundreds or even thousands. However, the present system may be configured to handle a number of repeat matches, including without the need for additional processing steps, such as when the number of matches is less than about 50, or less than about 40, or less than about 30, e.g. about 25 or less than about 20, for example, less than about 16 matches or less than about 10 or about 5 matches. However, in the event that the number of real hit matches returned exceeds this value, the system can be configured to implement a secondary hashing function, for example, using a secondary hash table.

[00224] Accordingly, rather than placing such seeds that are known to have an increased likelihood of redundancy in the primary hash table, such seeds may be placed in a secondary hash table, or a secondary region in the first hash table. Additionally, in some cases, an entry that does not carry any potential mapping multiplicity information for that seed but conveys a command to access a secondary hash table, such as an extension entry, may be placed in the primary hash table. For example, the extension entry may be an instruction, e.g., an instruction to increase the length of a primary, e.g., non-unique or repeating primer, to a longer one producing a more unique primer, e.g., by adding to it one or more additional adjacent bases, e.g., at the end (ends) of the seed to turn it into a longer seed sequence, which can then be hashed and looked up, for example in a secondary table.

[00225] The recording may be configured to inform or otherwise instruct about increasing a known excess seed by a predetermined amount, and may also contain instructions as to where and/or how to increase the seed. For example, since a hash table is usually calculated in advance, e.g., initially constructed from seeds generated from the reference genome(s), the table may know before creation which of the seeds generated from the reference genome are likely to occur multiple times (if any). . Accordingly, in various embodiments, it is possible to determine in advance which seeds may need to be moved to the secondary hash table. For example, when constructing a hash table index, the characteristics of the sequences of reference requests placed in the hash table as an index are known, so for each seed it can be determined whether it is likely to generate many hits, for example, from 10 to 10,000 hits .

[00226] More specifically, in various embodiments, an algorithm may be implemented to determine predicted matches of a given seed learned from reference reads and/or subject reads. If it is determined that any particular seed is likely to return many hits, a flag may be generated, such as an entry in, for example, a hash table cell, indicating that that particular seed has a high hit rate. In such an embodiment, the entry may also contain instructions to skip hashing of this seed and seeds like it, since it is impractical to carry out the specific number, for example 20-10,000 or more evaluations of such a seed necessary to accurately determine its meta-mapping. In such an embodiment, the primary hash function may be unable to accurately determine which position of all possible positions that this seed may match is the one to which the read actually aligns, and accordingly, for practical purposes, such a seed is not possible map accurately at this stage, the prime hash function is unlikely to return a useful result, such as a result indicating exactly where in the genome this seed actually corresponds.

[00227] In such an embodiment, the hashing function algorithm may be configured to compute the actions necessary to make the redundant seed more unique. For example, a secondary hash function can determine how much the seed needs to be extended, and in what order and position, to ensure that the seed is no longer redundant, but rather has a degree of uniqueness suitable for hashing. Accordingly, the entry may also include instructions to extend an excess seed (i.e., a seed present in excess), for example, by two, by four, by six, etc., but at one or both ends of the seed to obtain in advance a certain level of uniqueness. Thus. Thus, primers that initially appeared identical may be determined to be nonidentical.

[00228] For example, in some cases, a typical notation may include instructions that the duplicate primer should be extended by up to X odd or even bases, but, in some cases, extended by an even number of bases, such as from about 2 to 4, to from about 8 to 16, to about or about 64 or more bases, for example, an equal number of bases on each side. For example, in cases of extension of up to 64 bases, the entry may contain instructions to add 32 bases to each side of the primer. The number of bases by which the primer is extended can be varied and may be any suitable number depending on the design of the system. In some embodiments, a secondary hashing function can be used to determine how many bases to extend the primer to obtain a more reasonable number of matches. Accordingly, the extension may be of precise relative uniqueness, as in the case where there are only 1, 2, 3, or even up to 16, or 25, or 50 match positions for this pattern. In various embodiments, extending both ends of the primer equally may be useful to avoid the problem with reverse reads, but in various embodiments, the primer may be extended by one or more, unequal number of bases on both sides.

[00229] More specifically, as in one example, if a seed contains 28 bases, and an extension entry, such extension entry located in a primary hash table cell causes the hash function to extend the seed by, for example, 64 bases, then this the entry may also contain additional instructions on how to extend the primer, for example by adding 32 bases to each side of the primer. However, lengthening can be done in any convenient read position and can be symmetrical or asymmetrical. In some embodiments, the entry may contain instructions to the hashing function to symmetrically extend the primer, since in some cases such symmetrical extension may be more efficient, such as in the case of reverse complements discussed herein. In this embodiment, during extension, the same number of bases will be added to each of the opposite sides of the seed. However, in other embodiments, extension may be accomplished by adding an even or odd number of bases in an asymmetrical manner, and accordingly, the extension operation will not necessarily extend the seed by the same number of bases on each side. Typically the primary hash table is configured in such a way that it is not completely full. For example, it is advisable to use no more than 80% or 90% of its capacity. This is necessary for high search speed efficiency. In the case where there are a large number of collisions when hashing seeds into the same place when creating a table, the sorting engine will create a chain of references to other positions, and thanks to this, the search engine will be able to find the position assigned to the overloaded seed. The denser the table, the higher the number of collisions and the longer the chain required to find an actual match.

[00230] In various embodiments, such as when the original, redundant seed is 28 bases long, the record transmits an instruction to extend it, for example, from 18 to 32 to 64 bases, for example, on each of the opposite sides of the seed, digital seed representation, the digital representation of the seed can be approximately 64 bases × 2 bits per base = 128 bits. Accordingly, depending on the implementation of the mapping module, this size may be too large to handle in the primary hash table. Accordingly, in some embodiments, in order to cope with the need for such extensive processing, in some embodiments, the secondary hashing module may be configured to store information associated with larger size seeds. Since the number of seeds requiring extension is a fraction of the total number of seeds, the secondary hash table may be smaller than the primary hash table. However, in other embodiments, eg to reduce the need for modulus calculations, eg to save bits, known redundant portions of a sequence, eg a primary sequence, can be replaced by pre-selected variables, eg a predetermined sequence length. In this embodiment, since the redundant sequence is already known and identified, there is no need to fully represent it digitally. Instead, in various embodiments, all one has to do is replace the known, redundant sequence with a known variable sequence, and all one actually looks for is extended portions, such as tails, that have been added to one side of the sequence because they are the only non-redundant and new parts of the original sequences. Accordingly, in some embodiments, the primary sequence can be replaced with a shorter identifier code (such as a 24-bit proxy instead of a 56-bit representation), and then extension bases can be added to the proxy, such as a 36-bit extension (for example, making 60 bits) , which can then be placed in the primary table extension record. In this way, the complexity associated with reads that are too short and/or too long can be minimized and the advantage of only needing to maintain one or a small number of reads in the hash table can be achieved.

[00231] As stated above, the hash function described above may be implemented in software and/or hardware. The advantage of implementing a hashing module in hardware is the ability to speed up processes and, accordingly, carry them out much faster. For example, in cases where software may include various instructions for performing one or more of various functions, implementation of such instructions often requires storing and/or recalling and/or interpreting data and instructions, for example, before execution. However, as noted above and described in more detail herein, the chip can be hardware-wired to perform these functions without the need to invoke, interpret and/or implement one or more of the sequence of instructions. Instead, the chip may be configured (mounted) to directly perform such functions. Accordingly, in various aspects, the present invention relates to a special-purpose mounted device that can be configured such that part or all of the hashing module described above can be implemented by one or more network circuits, such as on-chip integrated circuits, e.g., FPGA, ASIC or structured ASIC.

[00232] For example, in various embodiments, a hash table index may be created and the hash function executed on the chip, and in other embodiments, the hash table index is generated off-chip, such as by software running on the host processor, but after generation it is loaded onto the chip and used, for example, during the operation of the hashing module. In some embodiments, the chip may have any suitable number of gigabytes, such as 8 gigabytes, such as 16 gigabytes, such as 32 gigabytes, such as 64 gigabytes, such as about 128 gigabytes. In various embodiments, the chip may be configured to perform various hashing module processes using only a portion or all of the memory resources. For example, in cases where a regular reference genome can be constructed, most of the memory can be devoted to storing the hash index of the reference, and/or for storing reads, and/or in order to reserve space for other functional modules used, as in the case , when 16 gigabytes are allocated for storing reads, 8 gigabytes can be allocated for hash index storage, and another 8 gigabytes can be allocated for other processing functions. In another example, if 32 gigabytes are allocated for storing reads, 26 gigabytes can be allocated for storing the primary hash table, 2.5 gigabytes can be allocated for storing the secondary table, and 1.5 gigabytes can be allocated for storing the reference genome.

[00233] In some implementations, the secondary hash table may be constructed such that its digital representation is larger than the primary hash table. For example, in various embodiments, the primary hash table may be configured to store hash entries of 8 bytes each, 8 entries per hash container, for a total of 64 bytes per container, and the secondary hash table may be configured to store 16 hashes. records, which gives a total of 128 bytes per container. For each hash entry containing redundant hash bits, the same hash key bits, the positions of possible matches in the reference genome are returned. Accordingly, for the primary hash table, up to 8 positions can be defined. Up to 16 positions can be defined for a secondary hash table.

[00234] Whether implemented in hardware or software, it may be useful to structure the hash table to avoid collisions. For example, there may be multiple seeds that, due to various system artifacts, will tend to insert themselves into the hash table at the same location, regardless of whether there is a match or not. Often, collisions can be partially avoided by structuring the hash table. Accordingly, in various embodiments, the hash table may be structured to avoid collisions, and accordingly, may be configured to include one or more virtual hash containers.

[00235] In various embodiments, a hash table can be a structure with a representation in 8-byte, 16-byte, 32-byte, 64-byte, 128-byte, or the like. But in various implementation examples, it may be useful to represent the hash table in a 64-byte format. This can be useful, for example, in cases where the hash function will be accessing memory such as DRAM (Dynamic Random Access Memory), such as in a standard DIMM or SODIMM form factor, as is the case where the minimum packet size is typically 64 bytes. In this embodiment, the processor design for accessing this storage device will be such that the number of bytes required to form a container in the hash table is also 64, and accordingly, maximum efficiency can be realized. However, if the table is structured in a 32-byte format, this will not be sufficient, since approximately half of the bytes delivered in the packet will contain information that the processor does not need. This will reduce the efficiency of byte delivery by half. Conversely, if the number of bytes used to form a container in the hash table is several times the minimum packet size, such as 128, efficiency will not suffer since the processor actually needs all the information returned in a single access. Accordingly, in cases where the optimal size of a memory access burst is a certain size, such as 64 bytes, the hash table can be designed to optimize the use of burst size, as is the case when the number of bytes allocated to represent segments in hash table, processed by the mapping function, for example, 64 bytes, the same as the size of the memory packet. Accordingly, when memory bandwidth is a limiting factor, the hash table can be structured to make optimal use of these limitations.

[00236] Further, it is also noted that although a record can fit into 8 bytes, the hash function can be designed so that 8 bytes are not read from the table to process a single record, as this may be inefficient. Instead, you can read all 8 records in the container, you can read them simultaneously, or part of them. This can be useful for optimizing the processing speed of the system, since, given the architecture described above, it will take the same amount of time to process all 8 records as it would to process just 1 record. Accordingly, in some embodiments, the mapping module may include a hash table, which may itself include multiple subpartitions, such as virtual partitions or containers, where each container may have 1 or more slots, such as 8 slots, such that it can accommodate one or more entries to check for collisions. However, under some conditions, specified one or more containers may become completely full of entries, so a means may be provided to sort additional entries into other containers and write information to the original container indicating what the hash table lookup mechanism should look for next. to detect a match.

[00237] Accordingly, in some cases it may also be useful to employ one or more additional methods, for example, for collision monitoring, and one such method may include one or more of linear probing and/or hash chaining. For example, if it is not known what exactly is being searched for in a hash table or portion thereof, such as in one container of the hash table, and a particular container is full, then the hash search function can be configured such that if one container, and when searching in If the required entry is not found, the function can be directed to the next container, for example, +1 container, and this container can in turn be checked. This way, when searching for a specific entry, all containers can be checked. Accordingly, such a search can be performed by looking through one container after another until it becomes clear that the entry is not found, such as when an empty slot is detected in at least one of the containers. In particular, if each container is filled sequentially and each container is searched in turn according to the order in which it was filled, then if an empty slot is found, as in the case of sequentially checking containers for a particular record, the empty slot may indicate that such record does not exist, because if it existed, it would have to be placed in this empty slot, if not in the previous containers.

[00238] More specifically, in cases where 64 bytes are allocated for storage in a hash container where 8 entries are stored, after receiving the invoked container, the mapping processor can work with all 8 entries simultaneously to determine which ones match and which do not . For example, when performing a search, such as searching a primer from a read obtained from sequenced DNA from a sample against a primer generated from a reference genome, the digital representation of the primer can be compared to the primers in all, for example, 8, records to find a match. This can lead to several results. A direct match may be found. A sample primer may be placed in a hash table and, in some cases, a match will not be found, for example because it is not exactly identical to any corresponding primer in the reference, for example due to a hardware error or sequencing error in relation to the seed or read from which it was generated, or because the genetic sequence of that individual differs from the reference genome. Or the seed may be placed in a hash table and multiple matches may be returned, as in cases where the pattern seed matches 2, 3, 5, 10, 15, 20, or more places in the table. In this embodiment, multiple entries can be returned, all pointing to different and distinct locations in the reference genome that this particular primer matches, the entries for these matches can either be in the same container, or multiple containers can be probed, resulting in which will return all meaningful results, such as matches.

[00239] In some embodiments, as is the case where perhaps space becomes a limiting factor in a hash table, such as in hash table containers, an additional mechanism can be implemented to resolve collisions and/or save space. For example, when space is limited, such as when there is a need to store more than 8 records in a container, or when desired, hash chain creations can be performed. Creating a hash chain may involve, for example, replacing an entry containing a particular position in the genomic sequence with a chain pointer entry that, instead of pointing to a position in the genome, points to some other address, such as the second container in the current hash table , for example, a primary hash table or a secondary hash table. The advantage of this method over the linear probing method is that it gives the hash search engine direct access to the container containing the tender entry, instead of sequentially checking the containers in order.

[00240] This method may be useful for certain system architectures. For example, hashable primary seeds, as in the case of a primary lookup, are placed at a specific position in the table, such as their original position, and when chained, these seeds are placed in a position that may be different from their original container. Accordingly, as stated above, the first portion of the seed in digital representation, for example, from about 26 to about 29 bits, can be hashed and searched in the first stage. In a second step, the remaining from about 27 to about 30 bits can be entered into the hash table, for example, as in a hash chain, as a means of checking the first pass. Accordingly, for any seed, its source address can be hashed in the first stage, and the secondary address bits can be used in the second, verification stage. Accordingly, the first portion of seeds may be placed in the primary entry position, and the second portion may be placed in the table at the secondary entry position. And, as stated above, in various embodiments, these two different recording positions may be arranged separately, as in a concatenated recording format. Accordingly, in any final chaining container, the chained record format may positionally separate elements/records that are intended to be accessed and probed by local primary containers, and records that are intended to be chained.

[00241] Such hash chains can be extended to various lengths. The advantage of this chaining is that in the case where one or more containers include one or more, for example, 2, 3, 4, 5, 6 or more empty record slots, these empty record slots can be used to store chain data hashing. Accordingly, in some embodiments, hash chaining may include starting with an empty slot in one container and chaining that slot to another slot in another container, where the two containers may be located at distant positions in the hash table. Care can also be taken to avoid confusion between records that are placed in a remote container as part of a hash chain, and "native" records that are hashed directly into the same container. As usual, the remaining about 27 to about 30 bits of the secondary access key are stored in records located remotely in a chained container, but because this chained container is located at a distance from the original hash container, confirming these about 27 to 30 will not be sufficient to guarantee that a matching hash record matches the original seed that reached that container through chaining, as opposed to another seed that reached the same container through direct access. (e.g., a check of about 27 to 30 could be a full verification if about 26 to 29 bits used to address the hash table are indirectly checked for proximity to the original hash container being accessed).

[00242] To avoid returning an incorrect hash entry without having to store all the hash keys in the entries, a positional system can be used in a chained container. Accordingly, the chained container must contain a record in a contiguous chain format that contains an additional chain pointer to continue the chain of containers if necessary; This chain continuation entry must be in the container slot after all native entries corresponding to direct hash access and before all remote entries related to the chain. During query processing, before following any chain pointer, any records occurring after the continuation record should be ignored, and after following any chain pointer, any record occurring after the continuation record should be ignored.

[00243] For example, in cases where containers are approximately 75%-85% full, it is possible to scan 8 containers and still find only 15-25 slots suitable for use, while the hash concatenating these slots can be found in 2, or 3, or 4 containers. In this case, the number of probing or concatenation rounds required to store the hash record matters because it affects the speed of the system. At runtime, if probing is required to find a record, it may be necessary to perform multiple hash lookups on, say, a 64-byte read in a container, which slows down the system. Hash chaining helps minimize the average number of accesses that have to be performed because a chained container, which can be selected from a wide area, can usually fit a larger excess of hash entries than a probe container, which must follow a strict order. Accordingly, a given number of redundant hash entries can usually be placed into a shorter sequence of concatenated containers than the necessarily sequential probing containers, which also limits the number of accesses required to locate these redundant entries in a query. However, probing retains value for fewer redundant hash entries because probing does not require sacrificing container slots in the chain pointer.

[00244] For example, once it has been determined where all the possible primer matches to the reference genome are, it is necessary to determine which of all those possible locations where a given read could possibly match is actually the correct position to which it is aligned. Accordingly, after mapping, there may be many positions where one or more reads are predicted to match the reference genome. Therefore, there may be multiple primers that purport to indicate exactly the same thing, for example they may match exactly the same position on the reference given the position of the primer in the read.

[00245] Therefore, for any given read, the actual alignment must be determined. This determination can be made in several different ways. In one case, all reads can be evaluated to determine their correct alignment to a reference genome based on the positions indicated by all the seeds from the read that returned position information during the hashed search. However, in various embodiments, it is possible to perform a seed chain filtering function on one or more seeds before performing the alignment.

[00246] For example, in some embodiments, primers associated with a given read that are believed to map to the same general location in the reference genome can be aggregated into a single strand that references the same general region. All seeds associated with a single read can be grouped into one or more seed chains, such that each seed is included in only one chain. It is these chains that then lead to read alignment with each specified position in the reference genome. Specifically, in various embodiments, all seeds that have the same supporting data indicating that they belong to the same general locations in the reference can be collected together to form one or more chains. Accordingly, primers that cluster together, or at least give the impression that they will be near each other in the reference genome, for example, within a certain band, will be grouped into a chain of primers, and primers outside the specified band will be converted into another chain seed

[00247] Once these different seeds are aggregated into one or more different seed strands, it can be determined which strand actually represents the correct strand to be aligned. This can be accomplished, at least in part, by using a filtering algorithm, which is a heuristic algorithm designed to eliminate weak seeds that have a high probability of not being true. In general, longer seed chains, relative to the length covered by the read, are more likely to be correct; Also, seed chains that include a larger number of seeds are more likely to be correct. In one example, a heuristic algorithm may be used whereby a relatively strong "upstream" seed chain, such as one that is long or contains many seeds, filters out a relatively weak "downstream" seed chain, such as one that is short or contains a small number of seeds.

[00248] In one embodiment, the length of the downstream chain defines a threshold length, for example, two times, so that it can be filtered by an upstream chain having at least the specified threshold length. In another embodiment, the number of seeds in the downstream chain defines a threshold number of seeds, for example, five times the number of seeds, so that it can be filtered by an upstream chain containing at least the specified threshold number of seeds. In another embodiment, the length of the downstream chain determines the threshold number of seeds, for example, the result of subtracting the length of the seeds from twice the number of seeds, so that it can be filtered by the upstream chain from at least the threshold number of seeds. In some embodiments, for example, if alignment of chimeric reads is required, only upstream seeds that substantially overlap downstream seeds within the read can filter them out.

[00249] This process screens out primers with a low probability of identifying a region of the reference genome where a high-quality read alignment can be found. This can therefore be beneficial because it reduces the number of alignments that need to be performed on each read, thereby increasing processing speed and saving time. Accordingly, this process can be used, in part, as a tuning characteristic that allows more detailed filtering of seed chains if higher speed is required, such as in high-speed mode, and if greater overall accuracy is required, such as in high-speed mode. accuracy, they perform less detailed filtering of seed chains, for example, they evaluate all seed chains.

[00250] In various embodiments, editing of the seeds may be carried out, for example, before the step of filtering the seed chains. For example, for each read, if all of the seeds of a given read are processed by the mapping function and none of them return hits, there may be a high probability of one or more errors in the read, such as a sequencer error. In such a case, an editing function, such as a single change editing process, such as a SNP editing process, can be performed on each seed, such as on seeds that did not return a match. For example, at position X, a single edit function might instruct the designated nucleotide to be replaced by one of 3 other nucleotides; determining whether the specified substitution, eg, SNP substitution, resulted in a hit, eg, a match. Said single change editing can be performed in the same manner at each position in the seed and/or at each seed in the read, for example, replacing each alternative base for each position in the seed. In addition, when a single change is introduced into one seed, the effects that change has on all other overlapping seeds can be determined based on the single change.

[00251] Such editing may also be performed for insertions, for example, adding one of four nucleotides to a given insertion position, X, and determining whether the hit results from the introduction of the specified substitution. Said editing can be performed for all four nucleotides and/or for all positions (X, X+1, X+2, X+3, etc.) in the primer, and/or for all primers in the reads. Such editing can also be performed for deletions, for example, when one of the four nucleotides at a given position, X, is deleted in the primer, and determining whether a hit results from introducing the specified deletion. This editing can then be repeated for all positions X+1, X+2, X+3, etc. Such editing, however, can significantly increase the amount of additional processing and time spent, for example, requiring a significant number of additional searches, for example, 2, or 3, or 4, or 5, or 10, or 50, or 100, or 200, etc. P. However, this additional amount of processing and time spent can be useful if such editing can be used to determine a valid hit, for example, a match obtained where there was previously no match. In such a case, it can usually then be determined that an error occurred, and then that the error has been corrected, thus saving the read.

[00252] An additional heuristic algorithm may also be used to determine whether an edit function should be performed by performing a calculation to determine the probability of receiving a hit when performing such an edit. Once a certain probability threshold is reached, such as 85% likelihood, such seed editing can be performed. For example, the system may generate various statistics for seed strings, such as calculating the number of high-frequency hits present and/or the number of seed strings that contain high-frequency hits, and thus determine whether editing of the seed strings is likely to be significant in determining matches . For example, in the case where a significant proportion of high-frequency hits are determined to be present, seed editing may be skipped because it is not likely to produce a set of sequences unique enough to obtain a hit from a reasonable number of hash table lookups, such as 100 or less, 50 or less, 40 or less, 30 or less, 20 or less, or 10 or less. Such statistics can be analyzed and a decision can then be made to perform or not perform seed editing. For example, if the statistics show that for any given read, half of the positions show no hits and the rest show high-frequency hits, seed editing may be appropriate because if no matches are returned, there may be an error, but if a large number of high-frequency matches are returned, performing seed editing may be simple make no sense.

[00253] The result of performing one or more of these mapping, filtering, and/or editing functions is a list of reads that includes, for each read, a list of all possible locations at which the read could match the reference genome. Therefore, the mapping function can be performed to quickly determine where reads from a FASTQ file obtained from a sequencer map onto a reference genome, for example, where different reads map in the complete genome. However, if there is an error in any of the reads or genetic variation, an exact match to the reference may not be obtained; and/or there may be multiple locations to which one or more reads are believed to match. Accordingly, it is necessary to determine where different reads are actually aligned relative to the genome as a whole.

[00254] Accordingly, after mapping and/or filtering and/or editing, location positions are determined for a large number of reads, with many individual reads having multiple location positions determined, and it is then necessary to determine which of all possible locations are actually true or the most likely locations to which different reads are aligned. Such an alignment can be performed using one or more algorithms, such as a dynamic programming algorithm, which finds matches for mapped reads to a reference genome and performs an alignment function on it.

[00255] An example alignment function compares one or more, for example, all reads to a reference, for example, by overlaying them in a graphical mode, for example, in a table, such as a virtual array or matrix, where the sequence of one of the reference genomes or mapped reads is placed on one dimension or axis, such as the horizontal axis, and the other is placed on the opposite dimensions or axis, such as the vertical axis. An imaginary scoring wavefront is then passed through the array to determine the alignment of the reads to the reference genome, for example by calculating alignment scores for each cell in the matrix.

[00256] The evaluation wavefront represents one or more, for example, all of the cells of the matrix, or a portion of the specified cells, which can be evaluated independently and/or simultaneously in accordance with the rules of dynamic programming applicable to the alignment algorithm, such as Smith-Waterman algorithms, and/or Needleman-Wunsch, and/or related algorithms. For example, if the upper left corner is taken as the beginning of the matrix (corresponding to the beginning of the read and/or the beginning of the reference window of the imaginary evaluation wave front), first only the upper left cell with coordinates (0,0) in the matrix can be evaluated, for example, a wave front of 1 cell ; then two cells can be evaluated in the direction to the right and down with coordinates (0,1) and (1,0), for example, a wave front of 2 cells; then three cells with coordinates (0,2), (1,1) and (2,0) can be evaluated, for example, a wave front of 3 cells. These wavefront examples may then expand diagonally along straight lines, from the lower left corner to the upper right corner, and the incremental movement of the wave front through the matrix occurs diagonally from the upper left corner to the lower right corner. Calculation of the alignment scores may be performed sequentially or in a different order, for example, by calculating all scores in the top row in a left-to-right direction, and then all scores in the next row in a left-to-right direction, and the like. In this approach, a rolling diagonal wavefront represents an optimal sequence of batches of estimates computed simultaneously or in parallel in a series of wavefront stages.

[00257] For example, in one embodiment, a reference genome window containing the segment to which the read has been mapped is placed on the horizontal axis and the read is placed on the vertical axis. In a similar way, an array or matrix is generated, for example, a virtual matrix, which allows you to compare the nucleotide at each position in the read with the nucleotide at each position in the reference window. As the wave front passes through the array, all potential ways to align the read with the reference window are considered, including whether changes in one sequence are required to ensure that the read matches the reference sequence, for example, by changing one or more nucleotides of the read to different nucleotides, or the insertion of one or more new nucleotides into one sequence, or the deletion of one or more nucleotides from one sequence.

[00258] An alignment score is generated reflecting the extent of changes required to achieve accurate alignment, which score and/or other associated data may be stored in specified array cells. Each array cell corresponds to the probability that a nucleotide at its position on the read axis is aligned with a nucleotide at its position on the reference axis, and the score generated for each cell represents the partial alignment ending with the cell's positions in the read and reference window. The highest score generated in any cell represents the best overall alignment of the read to the reference window. In various embodiments, the alignment can be global, where the entire read must be aligned with some part of the reference window, for example, using the Needleman-Wunsch algorithm or a similar algorithm; or, in other cases, the alignment may be local, where only part of the read can be aligned with part of the reference window, for example, using the Smith-Waterman algorithm or a similar algorithm.

[00259] The reference window can be of any suitable size. For example, since the length of a typical read can be from about 100 to about 1000 nucleotides, the length of the reference window accordingly, in some cases, can be from about 100 to 1000 nucleotides or more. However, in some cases, the length of the reads may be longer and/or the length of the reference window may be larger, for example, may be approximately 10,000, 25,000, 50,000, 75,000, 100,000, 200,000 nucleotides or more. It may be beneficial to slightly lengthen the reference window compared to the read, for example, including 32, or 64, or 128, or 200, or even 500 additional nucleotides in the reference window outside the segment of the reference genome to which the read was mapped, for example, so that It is possible to fully evaluate insertions and/or deletions near the ends of the read. For example, if only part of the read has been mapped to the reference segment, an additional extension may be applied to the reference window corresponding to the unmapped parts of the read, or extension by some factor, for example, by 10%, or 15%, or 20%, or 25%, or even 50%, or more, to ensure full alignment of the unmapped parts of the read volume with the reference window. However, in some cases the reference window length may be chosen to be less than the length of the reads, for example when a long part of the read is not mapped to a reference, for example more or less than 1000 nucleotides at one end of the read, for example to focus on the alignment of the mapped part.

[00260] The alignment wave front may have an unlimited length, or may be limited to any suitable fixed length, or have a variable length. For example, all cells along the full diagonal line of each wave front step, extending completely from one axis to another axis, can be evaluated. Alternatively, a limited length, for example 64 cells wide, can be estimated at each wavefront step, for example, by laying a diagonal strip of 64-cell-wide estimated cells in the matrix, leaving cells outside the specified strip unestimated. In some cases, it may not be necessary to calculate estimates far from the band around the true alignment path, and by calculating estimates only in a limited bandwidth using a fixed-length estimate wavefront, as described herein, a significant amount of work can be saved.

[00261] Accordingly, in various embodiments, the alignment function can be performed, for example, on data obtained from a mapping module. Accordingly, in various embodiments, the alignment function may form a module, such as an alignment module, which may form part of a system, such as a pipeline, that is used, for example, along with a mapping module, in the process of determining the actual complete genomic sequence of an individual or a portion thereof. For example, the output returned after executing a mapping function, e.g., from a mapping module, e.g., a list of possible mappings of one or more, or all reads to one or more positions in one or more reference genomes, can be used by the alignment function to determine the actual sequence alignment of the subject's sequenced DNA.

[00262] Such an alignment function can be useful in all cases because, as described above, often, for a variety of reasons, sequenced reads do not necessarily exactly match the reference genome. For example, one or more reads may contain an SNP (single nucleotide polymorphism), for example, the replacement of one nucleotide with another at a single position; there may be an "indel", an insertion or deletion of one or more bases along one or more read sequences not present in the reference genome; and/or there may be a sequencing error (eg, errors in sample preparation, and/or sequencer reads, and/or sequencer output, etc.) causing one or more of these apparent variations. Accordingly, when a read differs from the reference, for example, by the presence of an SNP or an indel, this may be due to the difference between the reference and the true DNA sequence in the sample, or the difference between the read and the true DNA sequence in the sample. The challenge is to correctly align the reads to the reference genome, given the fact that there are likely to be many differences between the two sequences.

[00263] Accordingly, in various embodiments, the input to the alignment function, for example, from a mapping function such as a prefix/suffix tree, or a Burrows-Wheeler transform, or a hash table, and/or a hash function, may be a list of possible options matches of one or more reads from one or more positions in one or more reference sequences. For example, any given read may match any number of positions in the reference genome, such as 1 location, or 16, or 32, or 64, or 100, or 500, or 1000 or more locations where a given read mapped to the genome. However, any given read was obtained, such as sequenced, from only one specific part of the genome. Accordingly, to find the true location from which a given read originates, an alignment function, such as a Smith-Waterman gap alignment, a Needleman-Wunsch alignment, etc., can be performed to determine the actual location of origin in the genome of one or more reads. for example, by comparing all possible locations where a match occurs and determining which of all possible options is the most likely location from which the read was sequenced, based on the highest location alignment score.

[00264] As stated, an algorithm is typically used to perform such an alignment function. For example, the Smith-Waterman and/or Needleman-Wunsch alignment algorithm can be used to align two or more sequences to each other. In this case, they can be used so that, for any given position where a read maps to a reference genome, the probabilities are determined that the mapping is actually performed at the position where the read originates. Typically, these algorithms are implemented in software, however, in various implementations such as those presented herein, one or more of these algorithms may be implemented in hardware, as described in more detail later herein.

[00265] In particular, the alignment function is used at least in part to align one or more, e.g., all reads to a reference genome despite the presence of one or more mismatched portions, e.g., SNPs, insertions, deletions, structural artifacts, and the like. ., to determine where specified reads are likely to fit correctly into the genome. For example, one or more reads are compared to a reference genome, and the best possible match of the read to the genome is determined, taking into account substitutions and/or indels and/or structural variants. However, in order to better determine which modified version of the read fits best into the reference genome, the putative changes must be taken into account, and accordingly a scoring function can also be performed.

[00266] For example, it is possible to perform an evaluation function, for example, as part of an overall alignment function, with the alignment module performing its function and introducing one or more changes to a sequence being compared to another sequence, for example, to achieve a better or better correspondence between them, for each change made to achieve a better alignment, a number is subtracted from the initial score, for example, either from the ideal score or from the zero initial score, so that when the alignment is performed, the assessment of this alignment is also determined, for example, when matches are detected, the score is increased, and a fine is imposed for each change made; thus, the best possible match for possible alignments can be determined, for example, by identifying from all possible modified reads the read that has the highest match score to the genome. Accordingly, in various embodiments, the alignment function may be configured to determine the best combination of changes to make to the read(s) to achieve the highest scoring alignment, which may accordingly be determined to be the correct or most likely alignment.

[00267] Accordingly, in view of the above, at least two purposes can be achieved by performing the alignment function. One of the goals is represented by a report of the best alignment, including the position in the reference genome and a description of the changes that are necessary for the read to match the reference segment at the specified position, and the other goal is represented by an assessment of the quality of the alignment. For example, in various embodiments, the output from the alignment module may be a "Compact Idiosyncratic Gapped Alignment Report", such as a CIGAR string, wherein said CIGAR output string is a report detailing all changes made to the reads to achieve their most appropriate alignments, for example, detailed alignment instructions indicating how the sequence of interest is actually aligned to the reference. The output of such a CIGAR string can be useful in subsequent processing steps to better determine that, for a given subject's genomic nucleotide sequence, the predicted variations compared to the reference genome are indeed true variations and not simply due to machine, software, or human error.

[00268] As discussed above, in various implementations, alignment is typically performed sequentially, with the algorithm receiving read sequence data, for example from a mapping module, belonging to the read, and one or more possible locations where the read could potentially be mapped to one or more reference genomes, and also receives genomic sequence data, for example, from one or more memories, relating to one or more positions in one or more reference genomes, to which the read can be mapped. Specifically, in various embodiments, the mapping module processes reads, for example from a FASTQ file, and maps each of them to one or more positions in the reference genome to which they may be aligned. The aligner then takes these predicted positions and uses them to align the reads to a reference genome, for example by constructing a virtual array against which the reads can be compared to the reference genome.

[00269] In performing this function, the aligner evaluates each mapped position for each individual read and, in particular, evaluates those reads that are mapped to many possible locations in the reference genome, and for each position evaluates the possibility that it is correct. It then compares the best estimates, such as the two best estimates, and makes a decision about where a particular read actually aligns. For example, when comparing the first and second best alignment scores, the aligner checks the difference between the scores, and if the difference between them is large, then the confidence rating that the position with the higher score is correct will be high. However, if the difference between them is small, for example, zero, then the assessment of the reliability of choosing one of them from two positions as the correct position from which the read was obtained is low; and additional processing may be useful to clearly identify the true location in the reference genome from which the read is derived. Accordingly, the aligner specifically looks for the largest difference between the first and second best confidence scores to decide that a given read maps to a given location in the reference genome. Ideally, the score of the best possible alignment is significantly higher than the score of the second best alignment for a given sequence.

[00270] There are many different ways to implement an alignment evaluation method, for example, you can evaluate each cell of an array or a subset of cells, for example, in accordance with the methods described herein. Typically, each alignment match corresponding to a diagonal step in the alignment matrix contributes to a positive score, for example +1 if the corresponding read and reference nucleotides; and a negative score, such as -4, if the two nucleotides do not match. Further, each deletion in the reference corresponding to a step in the horizontal direction in the alignment matrix contributes to a negative score, e.g., -7, and each insertion in the reference corresponding to a step in the vertical direction in the alignment matrix contributes to a negative score, e.g. -7.

[00271] In various embodiments, the scoring parameters for nucleotide matches, mismatches, nucleotide insertions and deletions can take on any of various positive or negative, or zero values. In various embodiments, these evaluation parameters may be modified based on available information. For example, in some embodiments, the affinity function penalizes alignment (insertion or deletion) gaps depending on the length of the gap, e.g., -7 for the first deleted (or inserted) nucleotide, but only -1 for each additional deleted (or inserted) nucleotide. , respectively, inserted) nucleotide in a continuous sequence. In various implementations, affine gap penalties may be implemented by dividing the gap penalties (for insertion or deletion) into two components, such as a gap opening penalty of, for example, -6 applied to the first step in the gap; and a gap extension penalty, such as -1, applied to each or subsequent steps in the gap. Affinity gap penalties can provide more accurate alignments, for example by allowing alignments involving long insertions or deletions to receive adequately high scores. Further, the cost of each horizontal movement may be the same or different, for example, the cost of a step may be the same, and/or, if there are gaps, such gaps may have a higher or lower cost, so the cost of horizontal leveler movements may be less than the cost of gaps. Accordingly, affine gap estimation can be implemented in various implementations, however this may be expensive in software and/or hardware since it typically requires multiple, e.g., 3 estimates, for each cell to be estimated, and, Accordingly, according to various embodiments, affine gap estimation is not implemented.

[00272] In various embodiments, the scoring parameters may also be sensitive to “base quality scores” corresponding to the nucleotides in the read. Some sequenced DNA read data, in formats such as FASTQ, may include a base quality score associated with each nucleotide, indicating the estimated probability of the nucleotide being incorrect, for example due to sequencing error. In some read data, base quality scores may indicate the likelihood of an insertion and/or deletion sequencing error at each position or an adjacent position, or additional quality scores may provide this information separately. More accurate alignment can accordingly be achieved by using scoring parameters including any or all of nucleotide match scores, nucleotide mismatch scores, gap (insertion and/or deletion) penalties, gap opening and/or gap extension penalties, varying in depending on the base quality score associated with the current nucleotide or current read position. For example, evaluation bonuses and/or penalties may be reduced if the evidence quality assessment indicates a high likelihood of a sequencing error or other error. The ground quality-sensitive evaluation may be implemented, for example, by using a fixed or configurable lookup table that is accessed using a ground quality score returning the appropriate scoring parameters.

[00273] When implemented in hardware on an integrated circuit such as an FPGA, ASIC, or structured ASIC, the evaluation wavefront may be implemented as a linear array of evaluation cells, for example, 16 cells, or 32 cells, or 64 cells, or 128 cells, or etc. Each of the evaluation cells can be constructed from digital gates in a wiring configuration to compute alignment estimates. Accordingly, for each wavefront stage, eg, each clock cycle, or some other fixed or variable unit of time, each of the evaluation cells, or portion of the cells, calculates the evaluation or evaluations required for a new cell in the virtual alignment matrix. Theoretically, different evaluation cells are considered to be at different positions in the alignment matrix corresponding to the evaluation wave front as described herein, for example, along a straight line extending from the lower left to the upper right of the matrix. As is well known in the field of digital logic device design, the physical evaluation cells and their constituent digital logic do not necessarily need to be physically located in this manner on the integrated circuit.

[00274] Accordingly, as the wave front sweeps step by step through the virtual alignment matrix, the imaginary evaluation cell positions update each cell accordingly, for example, speculatively "moving" a step to the right or, for example, a step down in the alignment matrix . All evaluation cells undergo the same relative imaginary movement, maintaining the order of the diagonal wave front. Each time a wave front moves to a new position, for example by a step in the vertical direction down or a step in the horizontal direction to the right in the matrix, the evaluation cells arrive at new imaginary positions and compute alignment scores for the cells of the virtual alignment matrix in which they are included. .

[00275] In such an implementation, adjacent scoring cells in a linear array are connected to exchange test nucleotides, reference nucleotides, and previously calculated alignment scores. The reference window nucleotides can be sequentially applied to one end of the wave front, for example, the top right score cell in a linear array, and can be sequentially shifted from there down along the wave front, such that at any given time a segment of reference nucleotides equal in length to the number of cells evaluation, is present in these cells, one of the successive nucleotides in each successive evaluation cell.

[00276] Accordingly, each time the wave front moves one step in the horizontal direction, the next reference nucleotide is fed into the upper right cell, and other reference nucleotides are moved down and to the left along the wave front. The indicated shift of reference nucleotides may be a real reflection of the imaginary movement of the wave front of the evaluation cells to the right in the alignment matrix. Accordingly, the nucleotides of the read can be sequentially fed to the opposite end of the wave front, for example, into the lower left evaluation cell in a linear array, and can be sequentially shifted from there upward along the wave front so that at any given time a segment of the nucleotides under study equal in length to the number of evaluation cells , was present in these cells, one of the consecutive nucleotides in each consecutive evaluation cell.

[00277] Similarly, each time the wave front moves one step in the vertical direction, the next nucleotide of interest is fed into the lower left cell, and other nucleotides of interest are moved up to the right along the wave front. The indicated shift of the studied nucleotides is a real reflection of the imaginary movement of the wave front of the evaluation cells downward in the alignment matrix. Accordingly, by issuing a command to shift reference nucleotides, one can move the wave front one step in the horizontal direction, and by issuing a command to shift the test nucleotides, one can move the wave front one step in the vertical direction. Accordingly, to obtain an overall diagonal movement of the wavefront, for example to follow a typical alignment of the test and reference sequences without insertions or deletions, the wavefront can be commanded alternately to move one step in the vertical and horizontal directions.

[00278] Accordingly, adjacent scoring cells in a linear array can be connected to exchange previously computed alignment scores. In various alignment estimation algorithms, such as Smith-Waterman or Needleman-Wunsch, or similar embodiments, the alignment score or scores in each cell of the virtual alignment matrix can be calculated based on previously computed scores in other matrix cells, for example, the three cells immediately to the left from the current cell, above the current cell, and up to the left diagonally from the current cell. When a score cell computes a new score or scores for a different matrix position, it shall retrieve such previously computed scores corresponding to such other matrix positions. Said previously computed scores may be obtained from a store of previously computed scores within the same cell and/or from a store of previously computed scores in one or two adjacent score cells in the linear array. This is because the three scoring-contributing positions in the virtual alignment matrix (directly left, top, and top left diagonally) could have been scored by either the current score cell or one of its adjacent score cells in the linear array.

[00279] For example, a cell immediately to the left in the matrix could be scored by the current evaluation cell if the most recent wavefront step was in the horizontal direction (to the right), or could be scored by an adjacent cell to the bottom left in a linear array if the most recent wavefront step was produced in the vertical direction (down). Likewise, a cell immediately above in the matrix could be scored by the current scoring cell if the most recent wavefront step was taken in the vertical direction (down), or could be scored by an adjacent cell on the top right in a linear array if the most recent wavefront step was taken in the horizontal direction (to the right). Likewise, the top-left diagonal cell in the matrix could be scored by the current scoring cell if the two most recent steps of the wave front were made in different directions, for example, down and then to the right; or right and then down; or could be scored by a top-right neighbor cell in a line array if the two most recent wavefront steps were both in the horizontal direction (to the right), or could be scored by a bottom-left neighbor cell in a line array if the two most recent wavefront steps were both in vertical direction (down).

[00280] Accordingly, given information about the directions of the last one or two steps of the wave front, an estimation cell can select the appropriate previously calculated estimates by accessing them within itself and/or in neighboring estimation cells by exploiting the connection between neighboring cells. In one embodiment, for rating cells located at two ends of the wave front, invalid, or zero, or minimum value ratings can be hardwired into the outer rating inputs so that they do not affect new rating calculations in those outermost cells.

[00281] By implementing a wavefront in this manner in a linear array of evaluation cells, with such a connection for shifting reference and test nucleotides across the array in opposite directions, for speculatively moving the wavefront stepwise in the vertical and horizontal direction, and a connection for accessing the estimates previously calculated by neighboring cells to calculate the alignment estimate or estimates at new positions of the cells of the virtual matrix into which the wave front enters, one can, accordingly, estimate a strip of cells in the virtual matrix, the width of the wave front, for example, by issuing commands to move the wave front sequentially step by step, so that it rolls around the matrix. To align the new read and the reference window, respectively, the wave front can start from inside the evaluation matrix, or, preferably, can gradually enter the evaluation matrix from the outside, starting, for example, from the left, or from above, or diagonally to the left and from the top left corner of the matrix .

[00282] For example, the wave front may begin with its top left evaluation cell located immediately to the left of the top left cell of the virtual matrix, and then the wave front may roll rightward into the matrix through a series of horizontal steps, evaluating a horizontal strip of cells in the top left region of the matrix . When the wavefront reaches the predicted alignment relationship between the reference and test sequences, or when a match is detected based on increasing alignment scores, the wavefront can begin to sweep diagonally down to the right by alternating steps in the vertical and horizontal directions, evaluating a diagonal stripe of cells in the middle of the matrix. When the lower left wavefront evaluation cell reaches the bottom boundary of the alignment matrix, the wavefront may begin to roll to the right again by successive steps in the horizontal direction, evaluating a horizontal strip of cells in the lower right region of the matrix, until some or all of the wavefront cells will go beyond the boundaries of the alignment matrix.

[00283] In one embodiment, increased efficiency of the alignment wavefront can be achieved by dividing its evaluation cells between two successive alignment operations. The next alignment matrix is set in advance because when the upper right portion of the wave front exits the lower right area of the current alignment matrix, it may immediately or after crossing a minimum gap such as one cell or three cells enter the upper right area of the next alignment matrix. Thus, the horizontal exit of a wave front from one alignment matrix may represent the same action as the horizontal entry of a wave front into the next alignment matrix. The implementation of this method may include the entry of reference and study bases of the next alignment into the evaluation cells, passing into the next alignment matrix, and may reduce the average length of time required for each alignment by the time of executing a number of wave front steps, practically equal to the number of alignment cells in the wave front , for example, 64 or 63 steps, or 61 steps, which may take, for example, 64 or 63 clock cycles, or 61 clock cycles.

[00284] The number of evaluation cells in implementing an alignment wavefront may be selected to balance various factors, including alignment accuracy, maximum insertion and deletion length, area, cost and power consumption of digital logic devices, equalizer logic clock speed, and performance general integrated circuit. A long wavefront is desirable for good alignment accuracy, particularly since a wavefront of N cells is capable of performing alignment over indels of approximately N nucleotides in length, or slightly shorter. However, using a longer wavefront requires more logic devices, resulting in more power consumption. Further, a longer wavefront can increase routing complexity and increase latency on the integrated circuit, resulting in lower maximum clock speeds that reduce the net equalizer performance. Moreover, if the integrated circuit is of limited size or power consumption, the use of a longer wavefront may require implementing fewer logic units on any portion of the IC, such as rendering fewer complete wavefronts, or other equalizer or mapper logic components, thereby reducing pure IC performance. In one particular embodiment, 64 evaluation cells in the wave front may provide an acceptable balance of these factors.

[00285] Accordingly, given a wave front width of X cells, for example, 64 evaluation cells, the width of the evaluation band in the alignment matrix will likewise be 64 cells (when measured diagonally). Matrix cells outside the specified band do not need to be processed or their scores calculated, as long as the optimal (best-scoring) alignment path in the matrix remains within the band being estimated. For a relatively small matrix, respectively, used to align relatively short reads, such as a read of 100 nucleotides or 250 nucleotides in length, this may be a safe assumption, for example if the wave front rolls exactly diagonally along the predicted aligned read position.

[00286] However, in some cases, for example in the case of a large alignment matrix used to align long reads, for example 1000, or 10,000, or 100,000 nucleotides in length, there may be a significant risk of indel accumulation causing the true alignment to deviate from the exact diagonal, totaling strong enough to leave the assessed band. In such cases, it may be useful to target the wave front so that the maximum set of estimates is located near the center of the wave front. Thus, as the wave front rolls, if the highest scores begin to move in one direction or another, for example, from left to right, the wave front moves to track said movement. For example, if the highest scores are observed in scoring cells located substantially above and to the right of the center of the wave front, said wave front may be driven some distance directly to the right by incremental steps in the horizontal direction until the highest scores return to an area near the center of the wave front.

[00287] Accordingly, an automatic guidance mechanism may be implemented in the wavefront control logic to determine a target position for guidance within the length of the wavefront, based on current and past estimates observed in the estimation wavefront cells, and to guide the wavefront to said target , if it is outside the center. More specifically, the position with the highest estimate at the last estimated position of the wave front can be used as a targeting target. In some cases this method is effective. However, in some cases the position with the highest score may not be a satisfactory targeting target. For example, with some combinations of alignment rating parameters, when a long indel occurs, when, accordingly, ratings begin to decline, a pattern can form along the wave front with two peaks of higher ratings and a trough of lower ratings between them, and as the indal continues, these two peaks continue to diverge.

[00288] Since it is not possible to easily determine whether a progressive event represents an insertion or a deletion, it is important to track the wavefront diagonally until the successful match begins again, either some distance in a direction to the right in the case of a deletion, or some distance in a downward direction in the case of an insertion . However, if two propagating estimate peaks are formed, one of them is likely to be slightly higher than the other, and may pull automatic guidance in that direction, causing the wave front to lose alignment if the actual indel was located in the other direction. A more robust method, accordingly, might be to subtract the delta value from the maximum observed wavefront score to determine the threshold score, identify the two outermost score cells that are at least equal to the specified threshold score, and use the midpoint between the specified outermost cells as a target for guidance This method provides a diagonal directional trend in the middle of the two-peak rating pattern. Other guidance criteria that serve to maintain higher scores near the center of the wave front can, however, easily be applied. When there is a delay in response between receiving an estimate from the wavefront of the evaluation cells and making the corresponding guidance decision, it may be appropriate to use hysteresis to compensate for guidance decisions made in the intervening period to avoid oscillating patterns of automatic wavefront guidance.

[00289] One or more such alignment procedures may be performed using any suitable alignment algorithm, such as the Needleman-Wunsch alignment algorithm and/or the Smith-Waterman alignment algorithm, which could be modified to accommodate the functional characteristics described herein. In general, both of these algorithms and similar algorithms generally function, in some cases, in a similar way. For example, as noted above, these alignment algorithms typically construct a virtual array in a similar manner, for example, in various embodiments, a horizontal top border may be configured to represent a genomic reference sequence that may be distributed across the top row of the array according to the pair composition grounds. Similarly, the vertical border may be configured to represent sequenced and mapped sequences of interest that have been arranged in a top-to-bottom order in the first column such that their nucleotide sequence order generally matches the nucleotide sequence of the reference to which they are mapped. The intermediate cells can then be filled with scores according to the probability that the relevant base of the sequence of interest at a given position is at the specified location relative to the reference. When performing this function, the capture strip can move diagonally across the matrix, filling the intermediate cells with estimates, and the probability of finding each base of the sequence of interest at a specified position can be determined.

[00290] In relation to the Needleman-Wunsch alignment function, which generates optimal global (or semi-global) alignments aligning the entire sequence of a read to some segment of a reference genome, wavefront guidance can be configured to generally sweep all the way from the topmost edge alignment matrix to the bottom edge. After the wave front has swept, the maximum score at the bottom edge of the alignment matrix (corresponding to the end of the read) is selected, and the alignment is tracked back to a cell at the top edge of the matrix (corresponding to the start of the read). In various cases described herein, reads may be of any length, may be of any size, and do not necessarily require comprehensive read parameters to determine how the alignment is performed, for example, in various embodiments, the length of the read may correspond to the length of the chromosome. In such a case, however, memory size and chromosome length may represent a limiting factor.

[00291] In relation to the Smith-Waterman algorithm, which generates optimal local alignments aligning all or part of the sequence of a read with some segment of a reference genome, the algorithm can be configured to find the best possible score based on a full or partial alignment of the read. Accordingly, in various embodiments, the band estimated by the wavefront may not extend to the top and/or bottom edges of the alignment matrix, for example, if a very long read contains only seeds in the middle of mapping to the reference genome, however, in general, the wavefront can still estimate from the top to the bottom edge of the matrix. Local alignment is typically achieved through two adjustments. First, the alignment scores are not allowed to fall below zero (or some other minimum level), and if the otherwise calculated cell score is negative, it is replaced with a zero score corresponding to the start of a new alignment. Second, the maximum alignment score obtained in any cell in the matrix, not necessarily along the bottom edge, is used as the end of the alignment. Alignment is tracked backwards from this maximum score up and to the left in the matrix to score zero, which is used as the starting position for local alignment, even if it is not located in the top row of the matrix.

[00292] Taking into account the above, there are several different possible paths through the virtual array. In various implementations, the wave front begins at the top left corner of the virtual array and moves down toward the maximum score identifiers. For example, the results of all possible alignments can be collected, processed, correlated, and scored to determine a maximum score. When the end of the boundary or the end of the array is reached, and/or the highest-scoring calculation has been made for all processed cells (eg, the overall highest score has been identified), backtracing can be performed to find the path that led to the specified highest score.

[00293] For example, a path that leads to a predicted maximum score may be identified, and once identified, a check may be performed to determine how the specified maximum score was obtained, for example, by backtracking along the alignment indicators with the best score, with tracking the path that led to the achievement of the identified maximum score, for example, calculated using the wave front score cells. This backward reconstruction or backtracking involves starting a movement at a certain maximum score, and working back through previous cells, tracing a path through the score cells that led to the maximum score, up to the edge of the table, and back to the starting boundary, e.g. array, or to zero in the case of local alignment.

[00294] During backtracking, after reaching a particular cell in the alignment matrix, the next backtracking step is to the adjacent cell immediately to the left or above, or diagonally up and to the left, that contributed to the best estimate selected to construct the estimate in the current cell. In this way, the evolution of the maximum score can be determined, thus revealing how the maximum score was achieved. Backtracking can be completed at a corner, or at an edge, or at a boundary, or can be completed at a zero estimate, such as the top left corner of an array. Accordingly, it is by means of such backtracking that the correct alignment is identified, thereby obtaining a CIGAR output string, for example, 3M, 2D, 8M, 4I, 16M, etc., indicating how a sample of the genomic sequence originating from an individual, or its part matches or otherwise aligns with the genomic sequence of the reference DNA.

[00295] Accordingly, once it has been determined where each read is mapped, and further determined where each read is aligned, for example, each relevant read is assigned a position and a quality score reflecting the likelihood that the specified position is the correct alignment, so that Since the nucleotide sequence of a subject's DNA is known, the order of the various reads and/or the genomic sequence of the subject's nucleic acid can be verified, for example, by performing a backtracking function that moves backwards up the array, determining the identity of each nucleic acid in the correct order in the genomic sequence sample . Therefore, in some aspects, the present invention relates to a backtracking function, such as one that is part of an alignment module that performs both alignment and backtracking functions, for example, a module that may be part of a pipeline of modules, such as a pipeline that is configured to receive raw sequence read data, for example in the form of an individual's genomic sample, and mapping and/or alignment of this data, which can then be classified.

[00296] To facilitate the backtracking operation, it is useful to store a score vector for each scored cell in an alignment matrix encoding the score selection decision. In the case of classic Smith-Waterman and/or Needleman-Wunsch estimation with linear gap penalties, the estimation vector can encode four possibilities, which may optionally be stored as a 2-bit integer from 0 to 3, for example: 0 = new alignment (zero selected grade); 1 = vertical alignment (score selected from cell above, modified by gap penalty in sequence); 2 = horizontal alignment (score selected from cell on left, modified by gap penalty in sequence); 3 = diagonal alignment (score selected from cell above and left, modified by nucleotide match or mismatch score). Optionally, it is possible to store the calculated score or scores for each matrix cell evaluated (beyond the maximum alignment score achieved, which is typically stored), however this is generally not necessary for backtracking and may require large amounts of memory. Performing backtracking then becomes following the evaluation vectors; When backtracking reaches a specific cell in the matrix, the next backtracking step determines the stored score vector for the specified cell, for example: 0 = end of backtracking; 1 = backward tracking in upward direction; 2 = reverse tracking in left direction; 3 = reverse tracking diagonally up and to the left.

[00297] Such evaluation vectors may be stored in a two-dimensional table arranged according to the dimensions of the alignment matrix, where only the elements corresponding to the cells evaluated by the wavefront are populated. Alternatively, to save memory, make it easier to record score vectors as they are generated, and make it easier to fit alignment matrices of different sizes, the score vectors can be stored in a table whose each row is sized to accommodate score vectors from a single wavefront of the score cells, e.g., 128 bits to store 64 2-bit wavefront estimate vectors, consisting of 64 cells, and the number of rows is equal to the maximum number of wavefront steps in the alignment operation.

[00298] Additionally, in this embodiment, it is possible to record the directions of various wavefront steps, for example, by storing an additional, for example, 129th bit in each row of the table, encoding, for example, with 0 vertical wavefront step preceding the specified edge position waves, and using a 1 - horizontal wave front step preceding the specified wave front position. This additional bit can be used in backtracking to keep track of which positions of the virtual score matrix the score vectors in each row of the table correspond to, so that the correct score vector can be retrieved after each successive backtracking step. If the backtracking step is vertical or horizontal, the next estimate vector should be extracted from the previous row of the table, however if the backtracking step is diagonal, the next estimate vector should be extracted from the previous two rows, since the wavefront had to take two steps to move from the estimate any cell to the score of the cell located diagonally to its lower right.

[00299] In the case of affine gap estimation, the estimation vector information can be expanded to, for example, 4 bits per cell being estimated. In addition to, for example, a 2-bit evaluation selection direction indicator, two 1-bit flags may be added, a vertical extension flag and a horizontal extension flag. According to affine gap estimation methods for Smith-Waterman or Needleman-Wunsch algorithms, or similar alignment algorithms, for each cell, in addition to the primary alignment score representing the best-scoring alignment ending in the specified cell, a "vertical" must be generated score" corresponding to the maximum alignment score reaching the specified cell at the last step in the vertical direction, and a "horizontal score" corresponding to the maximum alignment score reaching the specified cell at the last step in the horizontal direction; however, when calculating any of these three estimates, the vertical step into the cell can be calculated using the greater of two values: either the primary estimate from the cell above minus the penalty for opening a gap, or the vertical estimate from the cell above minus the penalty for extending the gap ; and the horizontal step into a cell can be calculated using the greater of either the raw estimate from the cell on the left minus the gap opening penalty, or the horizontal estimate from the cell on the left minus the gap extension penalty. In cases where vertical valuation minus the gap extension penalty is selected, the vertical extension flag in the valuation vector must be set, for example set to "1", and otherwise it must be cleared, for example set to "0". In cases where horizontal assessment minus gap extension penalty is selected, the horizontal extension flag in the assessment vector must be set, for example set to "1", and otherwise it must be cleared, for example set to "0". In backtracking, in the case of affine gap estimation, whenever backtracking takes a step in the vertical direction up from a given cell, if the specified cell has the vertical extension flag set in the estimation vector, the next backtracking step must also be vertical, regardless of the estimation vector for the cell on top. Similarly, whenever backtracking takes a step in the horizontal direction to the left of a given cell, if the horizontal extension flag in the score vector is set for that cell, the next backtrack step must also be horizontal, regardless of the score vector for the cell to the left.

[00300] Accordingly, such a table of estimation vectors, for example, 129 bits per row for 64 cells when using linear gap estimation, or 257 bits per row for 64 cells when using affine gap estimation, with some number (NR) of rows, is suitable to provide backtracking after completion of the alignment evaluation if the evaluation wavefront takes NR steps or less. For example, when aligning reads of 300 nucleotides, the number of wavefront steps required may always be less than 1024, so the table size may be 257 x 1024 bits, or approximately 32 kilobytes, which in many cases may represent a reasonable amount of local memory within the IC . However, if the alignment of very long reads, for example 100,000 nucleotides, is required, the memory requirements for the score vectors can be quite significant, for example 8 megabytes, which can be very expensive to include as local memory in the IC. To provide this support, the evaluation vector information may be recorded in a large capacity off-IC memory, such as DRAM, but in this case the bandwidth requirements, such as 257 bits per clock cycle per equalizer, may be prohibitive, which may constrain operation and performance. sharply reduce the performance of the equalizer.

[00301] Accordingly, it would be desirable to have a way of processing the score vectors prior to completion of the alignment so as to limit their storage requirements, for example, a way to perform incremental backtraces, generating incremental partial CIGAR rows, for example, from the initial parts of the history of the alignment score vectors, so that later the initial parts of the evaluation vectors could be removed. The difficulty is that backtracking, in principle, must start from the end of the alignment at a cell with a maximum score unknown until the end of the alignment evaluation, so any backtracking started before the alignment is complete may start from an incorrectly selected cell that is not on the potential the final path of optimal alignment.

[00302] Accordingly, a method is provided for performing incremental backtracking based on partial alignment information, for example, containing partial information about evaluation vectors for already evaluated alignment matrix cells. Based on the boundary of the alignment performed to date, for example, a specific estimated position of the wave front, back tracking begins from all cell positions on the specified boundary. Such backtracking from all cells on the edge can be performed sequentially or, preferably, particularly in the case of a hardware implementation, all backtracking is performed together. There is no need to highlight alignment notations, such as CIGAR strings specified by multiple backtracks; it is only necessary to determine which positions of the alignment matrix they pass through during backtracking. When implementing simultaneous backtracking from an evaluation boundary, a series of 1-bit registers corresponding to the number of alignment cells, all set to an initial value such as "1", can be used, which indicate whether any of the backtracking passes through the corresponding position. For each simultaneous backtracking step, the score vectors corresponding to all current "1" values in specified registers, for example, from a single row of the score vector table, can be examined to determine the next backtracking step corresponding to each "1" value in the specified registers leading to the next position for each value of "1" in the specified registers, for the next step of simultaneous backtracking.

[00303] It is important to note that it is quite likely that multiple "1" values in the registers will merge into common positions corresponding to multiple simultaneous backtracks merging on common backtrack paths. Once two or more simultaneous traces are merged, they remain merged indefinitely because they use the score vector information from the same cell from that point on. It has been observed, for both empirical and theoretical reasons, that it is highly likely that all simultaneous backtracks merge into a singular backtrack path in a relatively small number of backtrack steps, e.g., a number with a small multiple such as 8 greater than the number of evaluation cells in the front waves. For example, in the case of a 64-cell wavefront, there is a high probability that all backtracks from a given wavefront boundary will merge into a single backtrack path in no more than 512 backtrack steps. Alternatively, it is also possible and not uncommon to complete all tracebacks within a specified number, for example 512, of traceback steps.

[00304] Accordingly, multiple simultaneous backtracks can be performed from an estimation boundary, such as, for example, an estimated wavefront position far enough away that they all either terminate or merge into a single backtrack path, for example, in 512 backtrack steps or less. If they all merge into a singular backtracking path, then performing incremental backtracking based on partial alignment information is possible from the location in the evaluation matrix where they merge, or any distance back along the singular backtracking path. Additionally, start backtracking at the merge point, or any distance back, using conventional single backtracking techniques, including registering an appropriate alignment notation, such as a partial CIGAR string. The specified incremental backtrack and, for example, a partial CIGAR row shall be part of any eventual final backtrack, and, for example, a full CIGAR row, received after completion of the alignment, unless such final backtrack ends before reaching the evaluation boundary where it began simultaneous backtracking, since if it reaches the estimation boundary, it must follow one of the simultaneous backtracking paths and merge with the singular backtracking path that has been incrementally allocated by that point.

[00305] Accordingly, all score vectors for matrix regions corresponding to incrementally allocated backtracking, for example, in all table rows for wavefront positions preceding the start of the allocated singular backtracking, can be safely discarded. If, when performing a final backtrack from a cell with a maximum score, it ends before the score boundary is reached (or alternatively, if it ends before the start of the allocated singular backtrack is reached), the incremental alignment notation, such as the partial CIGAR string, may be discarded. If the final backtracking continues before the start of the singular backtracking allocation, its alignment notation, such as a CIGAR string, can then be appended to an incremental alignment notation, such as a partial CIGAR string.

[00306] Additionally, in very long alignment, the process of performing simultaneous backtracking from an estimation boundary, such as the estimated wavefront position, until all backtracking has completed or merged, followed by a singular backtracking highlighting the alignment notation, can be repeated several times, starting from different successive evaluation boundaries. Incremental alignment notation, such as a partial CIGAR string, from each successive incremental backtracking can then be appended to the accumulated previous alignment notations if the new simultaneous backtracking or singular backtracking does not end early; in this case, accumulated previous alignment notations may be discarded. A potential final backtracking similarly appends its own alignment notation to the most recent accumulated alignment notations to fully describe the backtracking, such as a CIGAR string.

[00307] Accordingly, the amount of memory for storing the evaluation vectors can be limited in this manner, provided that simultaneous backtraces are always merged with each other in a limited number of steps, for example, 512 steps. In rare cases where simultaneous backtraces do not merge and complete in a limited number of steps, various exceptional actions may be taken, including discarding the current alignment as erroneous or repeating it with a larger constraint or no constraint, likely different or traditional in a manner such as storing all score vectors for full alignment, for example, in external DRAM. In one embodiment, it may be appropriate to discard such an alignment as an error due to its extreme rarity, and it is even rarer for such an error alignment to be the best-scoring alignment for inclusion in the alignment report.

[00308] Optionally, the storage of the score vector can be divided, physically or logically, into a number of separate blocks, for example, 512 rows each, and the ending row in each block can be used as a score boundary to begin simultaneous backtracking. It may not necessarily be necessary to terminate or merge simultaneous backtracking within a single block, for example 512 steps. Optionally, if simultaneous backtracks are merged in a smaller number of steps, the merged backtrack can still continue throughout the entire block until the singular backtrack begins to be extracted in the previous block. Accordingly, after the evaluation vectors are completely written to block N and begin to be written to block N+1, simultaneous backtracking can begin in block N, followed by singular backtracking and allocation of the alignment notation in block N-1. If the rates of simultaneous backtracking, singular backtracking, and alignment evaluation are all similar or identical, and they can all be performed simultaneously, such as in parallel hardware in an IC, then the singular backtracking in block N-1 can occur simultaneously while populating the evaluation vectors block N+2, and when it is time to start filling block N+3, block N-1 can be released and returned to work.

[00309] Accordingly, with such an implementation, at least 4 blocks of evaluation vectors can be used, which can be used cyclically. Accordingly, the total storage capacity for the evaluation vectors for the equalizer module may include 4 blocks of 257×512 bits each, for example, or approximately 64 kilobytes. In one embodiment, if the current maximum alignment estimate corresponds to an earlier block than the current wavefront position, the specified block and the previous block may be retained rather than returned to operation so that the final backtracking can begin at the specified position if it remains the position with maximum rating; saving an additional 2 blocks in this manner shifts the minimum to, for example, 6 blocks. In another embodiment, additional blocks, for example 1 or 2 additional blocks, for example a total of 8 blocks, for example approximately 128 kilobytes. Accordingly, if such a limited number of blocks, for example 4 blocks or 8 blocks, are used in a round-robin manner, alignment and backtracking of reads of arbitrary length, for example 100,000 nucleotides, or an entire chromosome, is possible without the use of external memory for score vectors.

[00310] As described above, certain regions of DNA are represented by genes that encode proteins or functional RNA. Each gene exists on a single strand of a double-stranded DNA double helix, often as a series of exons (coding segments) separated by introns (non-coding segments). Some genes contain a single exon, but most contain several exons (separated by introns), and some contain hundreds of exons or thousands of exons. Exons are usually several hundred nucleotides in length, but they can be shorter, up to one nucleotide, or longer, up to tens, hundreds, or thousands of nucleotides. The length of introns is usually thousands of nucleotides, and some exceed a million nucleotides.

[00311] The gene can be transcribed by RNA polymerase enzymes into messenger RNA (mRNA) or other types of RNA. The directly produced RNA transcript is a single-stranded copy of the gene, except that the thymine (T) bases of the DNA are transcribed into the uracil (U) bases of RNA. However, copies of introns immediately after producing said copy tend to be spliced by splicingsomes, leaving copies of exons linked at "splice junctions" (which are not directly obvious thereafter). RNA splicing does not always occur the same way. Sometimes one or more exons are spliced, and sometimes the splice junctions do not fall at the most common intron/exon boundaries. Accordingly, a single gene can produce several different transcribed RNA segments through this process, sometimes called alternative splicing.

[00312] Spliced mRNA is transported (in eukaryotes) from the cell nucleus to the ribosome, which decodes it into protein, with each group of three RNA nucleotides (codon) encoding one amino acid. In this way, genes in DNA serve as the original instructions for making proteins.

[00313] Most RNA splicing occurs at stable exon/intron boundaries, which are characterized by typical sequence composition, particularly near the ends of introns. Specifically, the first two and last two bases of an intron, called an intron motif, have one of only 3 "canonical" intron motif sequences in the vast majority of cases (approximately 99.9%). The most common canonical intron motif is "GT/AG", which means that the first two bases of the intron are represented by "G" and "T", and the last two introns are represented by "A" and "G". The GT/AG motif occurs in approximately 98.8% of cases. Other canonical intronic motifs are GC/AG, which occurs in approximately 1.0% of cases, and AT/AC, which occurs in approximately 0.1% of cases. The indicated canonical motifs and their prevalence rates are quite stable in different species, but are not necessarily universal.

[00314] Not all genes are transcribed, and transcription of transcribed genes can occur at different rates. Many factors can influence whether a given gene will be transcribed into RNA, and how often. Some of these factors are inherited, some vary depending on the specialization of cells in different tissues, and some vary over time under the influence of environmental conditions or diseases. Accordingly, two cells with exactly the same DNA can produce quite different types and amounts of proteins and functional RNA. Because of this, sequencing (reading) the RNA present in one or more cells produces different information than sequencing DNA. A more complete picture of the state and activity of cells is provided by a combination of DNA sequencing and RNA sequencing.

[00315] Whole RNA transcriptome sequencing is typically performed by first selecting a target RNA, such as a protein-coding RNA, and then using enzymes with reverse transcriptase activity to convert the RNA segments back into complementary DNA (cDNA) strands. Said DNA can be amplified using polymerase chain reaction (PCR) and/or fragmented to obtain the desired sequence length distribution. The DNA fragments are then sequenced on a DNA sequencer, such as a next generation sequencer, using a "shotgun" method.

[00316] The resulting DNA reads are either reverse complementary or direct copies of the original RNA strands, except that the “U” is replaced by a “T” as described above. When preparing some libraries and performing some sequencing protocols, the orientation of the sequenced DNA strands relative to the original RNA may be preserved or labeled; but in conventional protocols, approximately 50% of the DNA sequenced is reverse complementary to the original RNA, with no direct indication of orientation (although there are indirect indications).

[00317] DNA reads from RNA sequencing (RNA-seq) protocols are different from those obtained from whole genome or whole exome DNA sequencing by other methods. First, except for contaminants, only transcribed RNA is sequenced, so non-coding DNA and inactive genes generally do not appear. Secondly, the number of sequenced reads corresponding to different genes is related to the biological transcription rates of these genes. Third, as a result of intron splicing, RNA-seq reads tend to skip intronic (non-coding) segments within genes.

[00318] RNA-seq reads are typically processed in a manner quite distinct from DNA reads. Although both types of reads are typically mapped to and aligned to a reference genome, the techniques for mapping and aligning DNA and RNA differ (see next section). After mapping and alignment, reads are usually sorted by reference mapping positions, in the case of both DNA and RNA. Double labeling, which may not necessarily be used in DNA processing, is not typically used for RNA-seq data.

[00319] The DNA reads are then typically processed using a variant detector to identify differences between the sample DNA and the reference genome. RNA-seq reads are not typically used to identify variants, although they are sometimes done. More often, aligned and sorted RNA reads are analyzed to determine which genes were expressed and in what relative amounts, or which of the various alternatively spliced transcripts were produced and in what relative amounts. This analysis typically involves counting the number of reads that are aligned to different genes, exons, etc., and may also involve assembly of transcripts (reference-based or de novo) to infer from relatively short RNA-seq reads that splicing is more likely to occur. long RNA transcripts from DNA.

[00320] Gene, exon, or transcript expression analysis is often extended to differential expression analysis, in which RNA sequencing data from multiple samples, often from two or more different classes (subpopulations or phenotypes), are compared to determine the extent to which the expression of said genes, exons, or transcripts differs. transcripts in different classes. This analysis may involve calculating the likelihood of the “null hypothesis” that the respective expression levels are the same across classes, as well as calculating the “fold change” of expression across samples, such as an 8-fold or 10-fold change, or changes with higher multiplicity.

[00321] In many DNA or RNA sequencing applications, reads are mapped to and aligned to a reference genome early in the processing. It is common to use a DNA-based reference genome when sequencing both DNA and RNA, containing a "T" rather than a "U", particularly since RNA-sequencing typically involves reverse transcription into cDNA prior to sequencing. In the case of RNA sequencing, as with whole exome sequencing for that matter, the reference genome could theoretically be limited to known coding regions or regions near the coding DNA. However, it is common practice to map and align the entire reference genome for the organism of interest.

[00322] The biggest difference required for a suitable RNA mapper/aligner is the ability to handle splice junctions. Because RNA-seq reads correspond to segments of transcribed and spliced RNA, the read typically crosses one or more splice junctions. When applied to a DNA-based reference genome, this means that the first part of the read comes from the first exon and should be mapped to it, and the second part of the read should be mapped to the second exon, etc. For example, in a read 100 bases long, the first 40 bases could be taken from an exon on chromosome 3 at offset 2,345,000, and the remaining 60 bases could be taken from another exon 100,000 bases away, starting on chromosome at offset 2,445,040. The alignment for such a read can be represented by the mapping position Chr3:2345000 and the CIGAR alignment string "40M100000N60M", in which "40M" and "60M" represent the parts aligned to the corresponding exons, and "100000N" represents the 100000 base intron, and the 100000 indicated reference bases are skipped when aligning the read. (Abstractly speaking, the specified CIGAR string can be considered equivalent to "40M100000D60M", where "100000D" represents a 100,000 base deletion from the reference, however putative spliced introns are commonly denoted as "N" as opposed to deletions of putative mutations or sequencing errors denoted as " D".)

[00323] The practical difference between CIGAR "N" (intron) and "D" (deletion) events is related to their typical length. Deletions are rarely longer than 50 bases and are thus successfully detected and accurately positioned when using the Smith-Waterman algorithm or similar algorithms for sequence alignment. Introns are often several thousand bases in length or even a million bases or more, and using Smith-Waterman-type aligners to detect such long alignment gaps is impractical. Therefore, initial exploration of splice junctions is more about “mapping” than “alignment.”

[00324] The problem with mapping is that each read may be divided into exon segments at unknown boundaries, and different exon segments are likely to be mapped to widely separated genomic locations that will need to be determined individually. Methods for mapping exon segments to their corresponding locations in the reference can be similar to methods for mapping an entire read to one reference segment, however, mapping with splicing (the first of these) is more complex, since each exon can be significantly shorter than the entire read, and, accordingly, contains significantly less information to guide the mapping device. Thus, a single exon may be as small as one (1) base in length, such as "G", and without additional information it would be impractical to attempt to determine where said single base should be mapped to with a potential range of intron lengths spanning millions of bases.

[00325] In addition to examining the mapping of two consecutive exon segments in a read, the splice junction between them must be precisely positioned, at least for some applications. While it may be clear that the first approximately 40 bases and the last approximately 60 bases of a 100-base-long read map to locations exactly 100,000 bases apart on chromosome 3, it is often much less clear exactly how many bases of a read map to each of those two. locations, or where exactly the splicing junction, the boundary between two exon segments, falls into the read. A valid CIGAR string might look, for example, not only as “40M100000N60M”, but also as “39M100000N61M” or “42M100000N58M”. Precise positioning of splice junctions is an “alignment” rather than a “mapping” operation.

[00326] A suitable RNA mapping device (mapper) can also successfully infer from which of the two DNA strands the read sequence was transcribed. In typical undirected RNA-seq protocols, a given read can be aligned either in direct or reverse complementary form to the reference (with or without splice junctions). In paired-end RNA-seq protocols, typically two paired reads are oriented “FR” (forward/reverse), such that a previously mapped paired read in the reference genome is forward oriented and the other paired read is reverse complementary. However, in typical undirected RNA-seq protocols, the specified mapping orientations do not determine which DNA strand carried the gene from which the RNA was transcribed for a given read, in part because cDNA amplification by PCR produces both orientations.

[00327] Finally, an RNA-based mapper can advantageously enhance an input database of “annotated” known splice boundaries. All common human genes have been studied in detail, for example, and the splicing boundaries of more common and less common RNA transcripts are annotated in genomic databases. This information is not 100% comprehensive; any individual samples will most likely have some "new" splicing not recorded in the databases. However, annotated splice boundaries can serve as useful guides to improve the accuracy of mapping and alignment of sequenced RNA. After mapping/aligning RNA-seq reads with or without annotated splice boundaries, an improved method is to detect the set of splice boundaries observed in the aligned reads, with some criteria such as the minimum number of alignments covering the splice boundary, and use this set empirically detected splice boundaries as annotated splice boundaries for the second RNA mapping/alignment run. This can increase sensitivity by using splice boundaries found in some reads to guide the mapping of other reads.

[00328] Seed mapping for RNA-seq reads is similar to DNA reads. The primary primer length is chosen, ideally, to be slightly longer than the base 4 logarithm of the reference genome size to ensure that primers are mapped approximately unambiguously, eg K=18 or K=21 for the entire human reference genome. A hash table is constructed, populated with some or all of the primers from the reference genome, a hash entry in the hash tables for each populated primer indicates its position and orientation in the reference. The hash table loaded into memory is made available to the mapper module hardware, such as DRAM modules on the FPGA board, wired to the FPGA pins, implementing the mapper module hardware.

[00329] The mapper module accepts RNA reads coming from an RNA or DNA sequencer (often reverse transcribed into cDNA before sequencing). From each read, the mapper extracts seeds of length K, ideally a floating window of many overlapping seeds of K bases, selected according to some pattern, such as starting at every base position or starting at every even position. The mapper accesses a hash table in memory for each primer, retrieving at least zero, one, or more positions in the reference genome that the primer matches. With respect to DNA mapping, primers can be dynamically expanded by repeatedly accessing a hash table with increasingly longer primers when it is necessary to reduce a large number of matching positions to an acceptably small number, such as 16 or fewer matches. Seed matches are combined into seed chains containing seeds matching the same orientation (forward or reverse match relative to the reference) along the same alignment diagonals.

[00330] For RNA-seq reads, it may be preferable to have an additional step performed by the mapper engine to refine the initial seed mapping by mapping the tethered short primer. For example, as shown in FIG. 1, RNA reads often cross one or more splice boundaries, and a primer that crosses a splice boundary usually cannot be mapped because its actual image in the reference is split between two locations. When a read contains a shorter exon than the initial length K of the primer, or a read overlaps a longer exon with fewer bases than K, then primer mapping may fail to determine the position in the corresponding reference for the specified exon. Even when a full or partial exon is slightly longer than K bases but shorter than the full read, it may be vulnerable to primer mapping error when it contains at least one edit relative to the reference, e.g. polymorphism (SNP, SNP) or insertion or deletion (indel) due to a mutation in the sample in relation to the reference or due to a sequencing error. For example, in FIG. Figure 1 shows an example of an error when mapping all exon segments with long (K bases) primers. For this reason, for good primer mapping sensitivity, it is preferable to turn to shorter primers that can fit into short exons or parts of exons extending beyond short reads, or between edits.

[00331] It may be somewhat impractical to consult a whole genome hash table for primers that are much shorter than the minimum length relative to base 4 logarithms of the reference genome size, since shorter primers will tend to match more locations. For example, with a reference of the entire human genome of approximately 3.1 billion bases in size, the logarithm to base 4 is approximately 15.8, and the minimum acceptable length of the primer accessed may be K=16 or 18, perhaps when K=21 is desired setting; It is not acceptable to refer to seeds with K=11 bases, since each 11-base model will match on average more than 700 reference locations.

[00332] However, after mapping the initial primers, for example with K=21 bases, it is possible to refine the seed mapping with anchored primers of shorter length, for example with L=11 bases. To map anchored seeds, the anchored hash table (which can be the same as the primary hash table or a separate hash table) is populated with L bases from the reference that are associated with specific regions of the reference, such as cells of some size. for example, 2 ¹⁶ =65,536 bases. Each reference region or bin is assigned a unique identifier, for example, its starting position in the reference genome divided by the size of the bin. Seeds with L bases in each reference cell are entered into a hash table of linked seeds using a hash key formed from the L bases of the seed and the cell identifier.

[00333] The mapper module may access a hash table of linked seeds for any given seed with L bases in any given cell using a hash key access similarly derived from the L bases of the seed and the cell ID. Through such access, only matches of a primer with L bases to a specific reference cell will be detected. Since a cell is significantly smaller than the entire reference genome, a short primer of L bases has enough information to often have only one match. For example, the base 4 logarithm of a cell of size 65536 is 8, so L=11 (or 10, 12, etc.) is an acceptable bound seed length for padding and reversal. As shown in FIG. 2, short (L bases) primers are more easily placed into a short exon, short exon overhangs, or edited exon segments such as SNPs.

[00334] The key to how anchored seed mapping works is that the mapper module's calls to the anchored seed hash table are guided by the seed mapping. Initial matches, such as those with K=21 bases, may not successfully map all exon segments of a read, but most likely map at least one exon segment of each read or its mate at the end of the read. Given at least one K base match within at least one exon segment in an RNA read or its mate, any other exon segments in the read that have not been successfully mapped by K base primers are most likely to match relatively closely in the reference genome .

[00335] For example, approximately 99% of human introns are shorter than 65,536 bases, so if one exon segment maps when primed with K bases at a given reference position, then other unmapped exon segments will most likely match within the same reference cell with 65 536 bases, or within an adjacent cell. As shown in FIG. 3, the search range can be defined, such as cell size, or cell size, or two cell sizes, and one or more reference cells within the search range of successfully mapped K-base seeds in a hash table of linked seeds can be accessed using L-base seeds. Thus, K-base seed matches serve as anchor points for local searches with shorter L-base primers. There is a high probability of finding additional matches to previously unmapped exon segments of the read. Thus, the sensitivity of primer mapping is increased for RNA reads.

[00336] In addition, there are various ways that the mapper module can use mapping of anchored short seeds. In one implementation, after the mapper accesses the initial K-base seed in the hash table and strings the matches together into seed chains, the mapper then retrieves the L-base seeds from the read and accesses them to nearby reference cells (within the selected range search for current seed chains) to find additional matches for shorter seeds with L bases, which the mapper module then combines into additional seed chains or adds to existing seed chains with the same alignment diagonals. In such an implementation, it is preferable for the anchored seed hash table to be the same as the primary hash table, or, for separate primary hash table and anchored seed hash table, to reside in the accessed memory at the same time . In both cases, in order for the K-base and L-base hash table(s) to fit into memory, approximately twice as much memory may be used, for example, 64 GB DRAM instead of 32 GB DRAM, or , alternatively, approximately half of the seeds of each length may be filled, for example, the seed filling density is 50% instead of 100%. To limit the number of required references to the hash table of anchored seeds, only the most promising seed strings can be used as anchor points and/or L-base seeds can be extracted from the read only from certain regions, for example, regions where K-base seeds are not present. were mapped successfully.

[00337] In another embodiment, mapping and/or alignment of a set of reads can be completed in the first run using only K base primers. The mapping/alignment results of each read can then be examined, for example by software outside the mapper module, to determine which reads require detailed mapping using short anchored seeds. One indicator that triggers refinement through anchored seeds is that the first run alignments are trimmed, particularly with trimming around or greater than the length L of the short seed. Another indicator that triggers drilling using anchored seeds could be a significant number of mismatches observed in the first-run alignments. Another indicator that triggers refinement by anchored seeds could be that both paired end-paired reads were not mapped successfully, mapped far apart, or with an unexpected relative orientation. Advantageously, if one read is selected for detailing using a short seed, then its paired end-conjugate read will also be selected. Advantageously, only a portion of the first pass alignments may require detailing using short seeds, such as 15% or 30%.

[00338] For each read in the subset to which short seed drilling is applied, one or more reference cells may be selected for search, for example, cells that overlap the search radius around the alignment results of the first run for the read and/or its associated read ( see FIGURE 3). A second mapping/alignment run can then be performed on the subset of reads selected for detailing. The second run can access the L base primers from each read to a hash table of anchored primers associated with one or more selected reference cells for each read. Typically, at least some of the reads selected for the second run will result in improved mapping/alignment, such as higher alignment scores; the results of the second run may be retained in cases where they are improved, or, in other cases, the results of the first run may be retained. If necessary, the primary hash table can be loaded into engine-available memory before the first mapping run and the bound seeds hash table can be loaded before the second mapping run, eliminating the need to place both hash tables (or one combined hash table) in memory at the same time , although both of them can be loaded simultaneously, or reducing the filling density of the reference seeds to accommodate both at the same time.

[00339] In some embodiments, the reference cells have a configurable size, the search radius is configurable, and the seed length (K) and the anchor seed length (L) are configurable. In other embodiments, the reference cell size is a power of two. The exemplary preferred settings for human RNA sequencing whole transcriptome processing are K=21, L=11, reference bin size 2 ¹⁶ =65536, and search radius 2 ¹⁴ =16384.

[00340] If annotated splice boundaries are passed to the mapper module, they can be used to improve mapping sensitivity. The list of annotated boundaries is loaded into memory, which is accessed by the mapper engine. Advantageously, the annotated boundaries may be formatted into a table readily accessible to the mapper module, e.g., a table with an entry for each, e.g., 1024-base, reference cell that either contains intron information with at least one endpoint in said cell, or points to a list (in space after the initial table) of multiple intron descriptors. Each intron descriptor indicates the reference position of both endpoints of the corresponding intron and can also carry additional information, such as which strand of DNA the intron gene is on, the intron motif, and an indicator of how often the splice boundary appears.

[00341] Once the seeds (initial seeds and/or short anchor seeds) have been mapped and seed chains have been formed, a table of annotated boundaries is accessed in rows corresponding to the reference regions covered by each seed chain, or regions near the ends of long seed chains. A list of introns with at least one nearby endpoint is obtained and compared with at least one accessed primer strand. Each intron is discarded if it is not a possible or probable splice boundary from the seed chain. In particular, by comparing the location of the intron endpoint in the reference with the endpoint of the primer chain in the reference and in the read, the actual location of the splice boundary in the read is calculated. If the specified valid location is outside the boundaries of the read or significantly overlaps the seed chain (for example, more than maxSpliceOlap = 16 within the endpoint of the seed chain) or is too far outside the boundaries of the seed chain in the read (for example, more than maxSpliceGap = 150 bases outside the seed chain), then the annotated boundary is discarded as least likely to be relevant.

[00342] Each remaining intron descriptor is considered a possible splice boundary at one end of the corresponding seed chain. This information is used in two ways. First, the opposite end of the intron in the reference is taken as the most likely location to which the adjacent part of the read should be mapped, even if that location was discovered by seed mapping. In effect, the most likely alignment diagonal at the opposite end of the intron is calculated exactly by adding the intron length to or subtracting the intron length from (depending on the orientation) the alignment diagonal at the corresponding end of the current seed chain. If the specified reference location diagonal alignment does not agree with the existing primer strand, then a new (pseudo) strand of primers is generated, starting from the reference location at the opposite end of the intron, and starting in the read at the corresponding position predicted by the calculated diagonal alignment. Thus, likely mapping locations of exon read segments are discovered without mapping primers within them, by inferring their locations in introns based on existing primer strings.

[00343] Second, the annotated intron information is used to establish a known relationship between two seed strands that represent adjacent exon segments in the read. The linkage information is recorded in one or both of the seed strand descriptors identifying the other strand to which it is linked by an annotated splice boundary. In addition, the exact position of the splice boundary is known (assuming the annotated splice boundary is correct), calculated by subtracting the annotated intron endpoints and the splice boundary alignment diagonals. The specified precise positioning of the splice boundary is also recorded in one or both of the seed chain descriptors.

[00344] If multiple annotated splice boundaries are found to be associated with a single seed strand, information about the connection and position of the splice boundary may be recorded in various ways. For example, each link between two chains may need to be written to only one of the two chains so that there is no conflict if, for example, it is always written at the "target" end of the link. A single seed chain descriptor may have space to store multiple links or have dynamic space for link information. Additionally, copies of existing seed chain descriptors can be made to contain additional communication information.

[00345] The viewing of the annotated splice boundary may advantageously be repeated. Starting from covering one seed chain, for example the first 1/3 of a read, an annotated splice boundary associated with a previously undetected reference location can be discovered, from which a new seed chain is formed. The annotated boundary table may be repeatedly accessed for the newly generated seed chain, possibly discovering that after the second 1/3 of the read there is another known boundary with a different undetected reference location. Advantageously, the annotated junction table entries may indicate the distance (travelling in the same direction as the annotated boundary) to reaching the nearest other annotated boundary within a transcript of the same gene, or overall. When the specified distance measured after the calculated splice boundary location in the read goes beyond the end of the read, there is no need to access the annotated boundary table again as nothing will be found.

[00346] Within the mapper module, seed matches with the same orientation (forward or inverse match to the reference) and the same alignment diagonals are combined into seed chains with the intention that a single alignment operation without or with a gap can then learn and evaluate the alignment between read and reference for each seed chain. A diagonal alignment can be represented as a diagonally oriented alignment path covered by a matching seed, within an alignment rectangle formed within a read sequence on one axis and a read sequence on the other axis; one representation as a whole can be calculated for forward alignments by subtracting the position of the seed in the read from its position in the reference, and for backward alignments by adding the position of the seed in the read with its position in the reference.

[00347] When a read exactly matches a segment of the reference, for example, positions from 0 to 100 in the read match positions from 1,200,000 to 1,200,100 in the reference, all seeds usually coincide along the same diagonal, for example, 1,200,000 - 0 = 1,200 100 - 100 = 1,200,000; a specific seed with 21 bases from bases 30 to 50 in the read will match bases from 1,200,030 to 1,200,050 in the reference, also on the same diagonal 1,200,030 - 30 = 1,200,000. Matches of seeds with the same orientation and diagonal are usually included in the same seed chain, but also seeds on slightly different alignment diagonals can be included in the same seed chain, for example seeds whose diagonals differ by no more than 20 or no more than 50, or more complex rules. Providing such tolerances for diagonal differences is useful since reads sometimes contain indels (insertions or deletions) relative to the reference and a gap alignment, such as the Smith-Waterman alignment in the aligner module can resolve and evaluate such indels for a single seed chain as long as the indels are not too large , for example, no more than 50 base insertions or deletions.

[00348] But RNA-seq reads often cross splice boundaries, in which the step from one base of the read to the next base of the read passes through the entire intron in the reference, which can be thousands of bases in length, or even more than a million bases in length. In such cases, primers on one side of the splice boundary in the read will map to a reference with significantly different alignment diagonals from those on the other side of the splice boundary; the integer difference of the diagonals is equal to the length of the missing intron, probably thousands or more than a million. Such seeds may not be accepted into the same seed chain, since the gap aligner cannot directly resolve such a long gap in the reference.

[00349] Thus, for mapping RNA, as opposed to DNA, it is expected that the correct alignment of a given read may contain multiple strands of primers, with each strand of primers corresponding to different exon segments in the read. Each candidate alignment may thus contain a sequence of multiple seed chains. The next step in the mapper module identifies such likely seed chain sequences, known herein as scaffolds.

[00350] Each scaffold, as a sequence of one or more primer strands, has a physical interpretation as a piecewise alignment of successive exon segments of the read with corresponding exon segments in the reference genome. Thus, each seed chain in a single scaffold should typically cover only a portion of the read, said portions extending from the start of the read toward the end of the read along the sequence of seed chains; and the corresponding reference segments of the primer strands must run in a fixed direction through the reference with intervening gaps corresponding to the expected intron lengths. Each scaffold will be run through an aligner module to resolve precise alignments and evaluate them, as well as the most likely alignment. But obtaining a list of scaffolds from a raw list of seed chains is difficult.

[00351] In practice, seed mapping for a single RNA mapped read can be obtained from a small number of seed strands to tens or more than hundreds of seed strands. With over a hundred seed chains, the number of potential seed chain sequences is extremely large. Thus, the problem is both obtaining an acceptably short list of scaffolds for processing in the aligner module, and determining said list of scaffolds from a given list of seed chains in an acceptable amount of time without slowing down the mapper module. In order to do this in an efficient manner, a recursive method is presented.

[00352] First, it is very useful to sort the seed strings in the order of the positions they cover in the read, for example, in ascending order of the starting positions of the seed strings in the read. Seed chains can be freely constructed in this order by accessing the hash table seeds from the start of the read to the end and forming them into seed chains in the same order. But if this is not the case, or if the order is disrupted by subsequent changes to the list of seed strings when mapping using anchored primers or searching for annotated splice boundaries, then the seed strings must be sorted before scaffold formation, for example, by "quicksort" or another sorting algorithm.

[00353] Next, rules are established by which one seed chain (B) can immediately follow another seed chain (A) in the same scaffold, establishing the relationship of seed chains from A to B. There is considerable flexibility in the rules, which can well To work, the rules must allow for probable seed-chain connections in the scaffold with the correct alignment, while excluding as many unlikely seed-chain connections as possible. There is a good working set of rules with various named parameters and good default values.

[00354] Criteria for seed string B to follow seed string A in the scaffold:

A and B have the same orientation

(Gap between A and B in the read) =: gap ≤ maxSpliceGap = 150

(Overlap between A and B in the read) =: olap ≤ maxSpliceOlap = 16

(The gap between the beginning of A and the beginning of B in the read) =: nhead ≥ olap + (olapAdj = 4)

(The gap between end A and end B in the read) =: tail ≥ olap + (olapAdj = 4)

(Reference gap A/B minus read gap A/B) =: intronLen ≥ minIntronLen = 20

(Reference interval A/B minus read interval A/B) =: intronLen ≤ maxIntronLen = 1000000

[00355] When annotated splice boundaries are used and an annotated link has been written between seed strands A and B, they can always follow each other.

[00356] The following is a recursive algorithm for generating multiple scaffolds:

Sort N seed chains by starting position in the read, if necessary

Loop c0 = 0 to N-1

Skip c0 if already used inside the scaffold

Initialize last = 0, scaf[0] = c0, start = c0 + 1, stack[0] = 1, stack[1] = 0, stackPos = 0

Loop until last ≥ 0

Loop c = start until N - 1

If the c chain can follow the scaf[last] chain:

scaf[++last] = c

stack[last] = 0 if stack[last] = c

Otherwise if last > stackPos and chain c can follow chain scaf[last-1]:

stack[last] = c

Output scaffold scaf[0 .. last]

Set stackPos = maximum to (0 .. last) with stack[stackPos] > 0

Set start = scaf[stackPos] + 1

Set last = stackPos - 1

[00357] Meanings of terms and variables in the algorithm:

“chain”: index 0..N-1 of the seed chain

scaf[] = scaffold being formed, each slot receives a 0..N-1 chain

c0 = first chain in the scaffold (slot 0)

last = final slot # (currently) in the scaffold

start = first chain in the search loop

stack[i] = highest numbered alternate chain for scaf[i], or 0 if missing. This represents the end point of searching for alternatives for scaf[i] after a recursive reservation.

stackPos = target scaffold slot intended to be replaced via recursive reservation. It is observed that when a backup occurs, stack[stackPos] is stacked and will not be updated until it is cleared.

[00358] The recursive search is performed in the physical logical part of the mapper module. Time may be available to execute the specified algorithm without significantly slowing down the module using hardware parallelization techniques. In particular, a group of seed chains for a given read may be buffered for scaffold-related logic for processing later in the processing pipeline chain, in parallel with the next read's seed-mapping and chain-forming logic processing.

[00359] Sometimes in practice recursion can develop, so it is necessary to limit it. A useful way to limit recursion, as well as limit the number of scaffolds obtained, is to filter smaller scaffolds as they are obtained. A useful method for filtering scaffolds is presented. For each scaffold, its read coverage is calculated as the number of read bases covered by one or more seed chains in the scaffold. Greater scaffold coverage most likely represents correct alignment. In particular, if the maximum read coverage among all generated scaffolds (so far) is detected, then scaffolds with a large coverage delta beyond the maximum coverage are less likely to represent the correct alignment.

[00360] In addition, for each scaffold, its reference genome coverage, the distance between the outermost bases in the reference of the first and last strand of primers in the scaffold, is calculated. Scaffolds with very large reference coverage are less likely to represent the correct alignment. Combining the two specified measurement results as follows is a powerful scaffold filter: filter_metric = (max_coverage - coverage) + floor(25 * (log ₂ (ref_span + 2 ¹³ ) - 13). The constants 25 and 13 should be configurable parameters: rna-filt- ratio = 25, and rna-span-log-min = 13. Scaffolds with many chains are filtered out, where the specified indicator exceeds a configurable threshold, for example rna-max-covg-gap = 150. A threshold of 200 significantly weakens the filter, and 100 significantly compacts.

[00361] The filter can be applied to completed or incomplete scaffolds obtained from the list of seed chains for a given read, by tracking or calculating the maximum coverage among all scaffolds, as well as scanning the list of scaffolds, and discarding those for which filter_metric > rna-max- covg-gap.

[00362] Such a filter can also be applied as a recursive cutoff in the middle of a cluster of scaffolds. As each new seed chain is added to the scaffold, the updated reference coverage is annotated, as well as the potential coverage, calculated by subtracting the coverage gaps within and before the specified seed chain from the read length. Using the specified partial coverage and potential coverage, if the metric triggers filtering, then any longer scaffold using the current candidate scaffold as a prefix will most likely be filtered, since the reference's coverage will only increase and the potential coverage will only decrease. Thus, all recursion that has the current candidate scaffold as a prefix can be skipped. Recursion pruning using a scaffold filter in this way can significantly reduce the recursion length to generate a set of scaffolds from a long list of seed chains.

[00363] The implementation of recursive scaffold searches can also be optimized. Recursive loops repeatedly scan parts of the list of seed chains, and the speed of the algorithm is thus significantly better when the list of seed chains is shorter. But in reality, it is not necessary to perform a recursive algorithm on the entire list of seed chains when some seed chains cannot form a scaffold with other seed chains. One optimization technique is to detect "isolated" primer strings that are farther than maxIntronLen (e.g., 1,000,000 bases) from any other primer string in the reference. Isolated seed chains can be automatically isolated as single-strand scaffolds and removed from the seed chain list before further scaffold formation, thereby shortening the list of seed chains scanned during the recursion process. Likewise, well-separated subsets of seed chains, such as subsets with each chromosome or subsets separated by more than maxIntronLen in the reference, can be detected, and recursive scaffolding can be performed separately on each such subset, resulting in reduced overall execution time.

[00364] The first step of the aligner module for each scaffold is to precisely locate each splice boundary, the boundary between two exon segments (represented by the two corresponding primer strands in the scaffold). This is called "stacking" exon segments together or stitching splice boundaries. More precise cross-linking is still needed after seed mapping and scaffold formation, since two consecutive seed chains by themselves may not provide a clear indication of where the correct boundary lies between them. For example, successive chains of primers may be separated by some distance in the read if the primers could not be mapped to one or both sides of the splice boundary at once; or successive chains of primers may overlap each other in a read, especially if the read sequence at the end of one exon matches the sequence at the beginning of the next exon. Even if successive strands of primers adjoin without a gap or overlap, this does not guarantee that the boundary between them lies at the correct position of the splice boundary.

[00365] Stitching splice boundaries in this manner is primarily an analysis to select the best position for stitching between successive exon segments in a read corresponding to the most likely position of the splice boundary. Two factors are useful in determining this. The first factor is to compare the read sequence with the left and right reference sequence at two exon segment mapping locations in the reference genome. This stitching position implies. that read bases left with respect to the suture are mapped to the left reference area, and read bases right with respect to the suture are mapped to the right reference area. When moving a potential stitching position from left to right in a read, the bases of the read change their mapping when they intersect from the right reference region to the left.

[00366] As shown in FIG. 4, the correct position of the splice boundary will most likely have a good match between the left part of the read and the left reference region, as well as between the right part of the read and the right reference region. The total number of mismatches (or SNPs) can be counted on both sides of a potential stitching position by comparison with the corresponding reference region; and cross-linking positions with fewer SNPs are most likely to be correct. A comparison of the left and right parts of the read for each stitching position is shown in FIG. 4.

[00367] SNP counting is effectively modeled by scanning crosslinking positions through a window of possible positions in the read, eg by overlapping each of two seed chains by no more than a certain distance, for example 48 bases. Said scanning is run in the hardware equalizer module, for example, at a speed of one position per cycle. Each time the scan moves one step, for example from left to right, only one read base changes its mapping from the right reference region to the left reference region. Thus, the left sequence comparison either gains one SNP or remains unchanged, and the right sequence comparison either loses one SNP or remains unchanged; and thus the direct number of SNPs changes to -1, 0 or +1. This gradual change in the number of SNPs for each step can be calculated by comparing one read base (the one crossed in the stitching position step) with two reference bases. If such a gradual change in the number of SNPs is added up in steps from left to right, then the current sum can be taken as a relative estimate, where the minimum estimate is the best. Equivalently, each matching base can be assigned a positive match penalty, and each non-matching base a negative mismatch penalty; and the sum of the incremental changes in the score should be maximum for the best stitching position.

[00368] Another factor implied by each cross-linking position is the intron motif. An intron motif is defined as the first two bases and last two bases of the missing reference segment, or intron. Equivalently, the motif for any potential cross-linking position is formed from the first two bases after the left reference region and the last two bases before the right reference region, as shown in FIG. 4. This putative intron motif depends on the position of the crosslink and usually changes when scanning the position of the crosslink in a window of possible crosslinking positions. Some “canonical” intronic motifs occur significantly more frequently than others during natural RNA splicing. A stitch position that matches a canonical intron motif is most likely to be the correct splice boundary position, especially if it is one of the most common canonical motifs.

[00369] A table of the three canonical intron motifs in human RNA is shown in TABLE I. For each motif, its inverse mapping is also shown, since in most RNA sequencing protocols reads can be mapped either forward or backward to the transcribed gene strand, so that although only the "forward" canonical motif arises in the original gene strand read, its reverse match can arise in RNA-seq reads. For each canonical motif and for the remaining class or non-canonical motifs, the average frequency in human RNA splicing is shown along with a sample scoring penalty that can be used, for example, with base matches with a score of +1 and base mismatches with a score of -4.

TABLE I Intron motif Reverse Match Approximate Frequency Fine assessment GT/AG CT/AC 98.73% 0 GC/AG CT/GC 1.03% 10 AT/AC GT/AT 0.11% 15 250 non-canonical motifs 0.13% 25

[00370] The splicing cross-linking module accordingly scans potential cross-linking positions in a window of possible cross-linking locations, for example from left to right, summing up the gradual change in score due to bases changing which region they map to, as well as subtracting at each potential cross-linking position the intron motif penalty according to intron motif observed immediately after the left reference region and immediately before the right reference region, and selects the position with the highest score for stitching.

[00371] In addition, some special results may be reviewed and evaluated. Stitching at the left edge or right edge of the window of possible stitching positions can be considered a stitching error and is most likely to occur when one of the two reference regions is not the correct mapping position for a segment of an exon read, for example when an annotated splice boundary was followed but turns out to be correct for a given rida. Left or right stitching may advantageously be given a bonus score of, for example, 25 on the same exemplary scoring scale, so that for successful stitching there must be significant evidence of the correct splicing boundary.

[00372] Additionally, if an annotated splice boundary is identified as linking two seed strands to be stitched, the annotated boundary is at a known position in the window of possible stitching positions. As an option, the stitching operation can be skipped, simply accepting the known position of the annotated boundary. As another option, the stitching operation can be performed, but the known position of the annotated boundary can be given a bonus score and/or automatically assigned the best available intron motif penalty, or a penalty of zero. As another option, instead of an intron motif penalty, a known position of an annotated boundary can be given a bonus score or penalty associated with the observed commonness or rarity of that splice boundary that is annotated in the annotation database. If a known position of an annotated splice boundary is selected for stitching, then the stitching boundary may be marked as consistent with the annotated boundary, so that this fact can be reported if the specified splice boundary occurs in the output read alignment.

[00373] Selected stitching positions can be annotated into scaffolds in various ways. In a preferred embodiment, the constituent seed chains of the scaffold are edited to start and end immediately adjacent to the selected crosslinking positions.

In addition, it is preferable for the aligner module to perform large edits to scaffolds based on the stitching results in some circumstances. If cross-linking fails, the scaffold may be truncated or split into two scaffolds at the point of error. Additionally, cross-linking between non-adjacent seed chains in a scaffold may be attempted, eg by skipping one seed chain. For example, for a scaffold containing primer strands 1, 2, 3, and 4, splice boundary stitching should naturally occur between strand pairs (1,2), (2,3), and (3,4); but, in addition, a stitching can be attempted between pairs of chains (1,3) and (2,4). If cross-linking 1 to 3 has a better score than cross-linking 1 to 2 followed by cross-linking 2 to 3, then seed chain 2 may be discarded from the scaffold.

[00375] Having determined the exact positions of the splice boundary in a given scaffold by stitching, the corresponding completed alignments and alignment scores for each scaffold can be determined through the use of a spaceless aligner module or a gapped aligner (eg, Smith-Waterman). For a scaffold with only one strand of primers, this is not much different from the alignment for DNA reads and the same hardware modules and methods can be used. For a scaffold with multiple primer strands, some additional methods are required to obtain a complete, possibly spliced (containing intronic operations) alignment.

[00376] One way in which completed spliced alignments can be determined is to separately align each exon read segment corresponding to each primer strand in the scaffold with its corresponding reference segment, using an intermediate-free aligner and/or a gap aligner. This approach has disadvantages when local (i.e., possibly trimmed) alignments are required. If alignments of individual exon segments are obtained without trimming, they are not easily assembled together into a complete spliced alignment. If alignments of individual exon segments are obtained without trimming, then their alignment estimates can be examined to determine whether the best overall local alignment will trim one or more exon segments, but the appropriate cutoff at arbitrary locations within exon segments is not easy to determine. Thus, it can be expensive to obtain trimmed and untrimmed versions of the alignment of each exon segment to overcome these difficulties.

[00377] One method for determining completed spliced alignments for a multi-strand scaffold involves concatenating exon segments together before alignment. Each aligner module - intermediate-free or with a gap - can supply two nucleotide sequences for alignment, the considered sequence (read) and the reference sequence. The combined sequence in question may simply be the entire read, representing the concatenation of its exon segments, optionally trimming the start or end of the read if the first or last exon segment does not extend to the start or end of the read.

[00378] The combined reference sequence is obtained by retrieving a reference genome segment representing a mapped image of each exon segment primer strand and combining said reference segments together. It should be noted that for a given exon segment (seed chain), its reference segment may differ in length from its read segment in the case where the leftmost primers in the primer chain fall on a slightly different alignment diagonal than the rightmost primers; for example, the seeds in the seed chain assume the presence of an indel. In this case, a spacer leveler should be used.

[00379] In addition, for alignment with a gap, the first and last exon segments of the reference sequence may extend outward, for example, through the first exon segment with 50 preceding reference bases and the last exon segment and 50 subsequent reference bases to provide space for deletions within the first and the last exon segments.

[00380] By combining scaffold exon segments from a read and a reference, it is possible to feed a single combined reference sequence and a single combined reference sequence into the aligner module, which can then function in essentially the same way for spliced RNA alignments as for (RNA) alignments. or DNA) without splicing. However, some other modification will provide an advantage. First, to determine appropriate trimming of local alignments at any position within any exon segment, some scoring penalties (or bonuses) can be applied to each splice boundary that the alignment crosses. In one embodiment, the scoring penalty for each splice boundary relates to its intron motif and annotated splice boundary status, and may be the same scoring penalty used when stitching the splice boundaries.

[00381] Accordingly, an unannotated splice boundary with a rare or non-canonical intronic motif may have a large associated score penalty and most likely have one or more of all exon segments trimmed from the splice alignment to exclude such unlikely splice boundary until it occurs a sufficient number of sequence matches on both sides of the splice boundary to serve as strong evidence that the splice boundary is indeed present, by overcoming its scoring penalty. In a preferred embodiment, each of the combined query and reference sequences has a decoy inserted between successive exon segments, and the appropriate scoring penalty for each splice boundary is attached to its corresponding decoy. This allows splice boundary estimation penalties to be enabled without specialized logic, and provides room for possible alignment trimming on either side of the splice boundary spurious base.

[00382] Additionally, in various examples, for a gap alignment such as Smith-Waterman, it may be advantageous to force portions of the alignment to pass through predefined splice boundaries. In other words, no alignment path should move from one reference exon segment to the next without simultaneously moving from the corresponding reference exon segment to the next. One reason for this limitation is that only a properly synchronized splice boundary will have a proper score based on the intron motif determined during splice boundary stitching.

[00383] Another reason is that the combined reference sequence was generated using the precise boundaries of the reference exon segment corresponding to the selected stitching position of each splice boundary, so that there are no additional reference bases for the spaced aligner to freely adjust the splice boundaries. Additionally, in order to avoid alignments that are difficult to interpret (for example, such as CIGAR strings with 'I' or 'D' operations adjacent to 'N' operations), it is advisable to require that at least one radix in question and a radix the reference base before each included splice boundary was aligned "diagonally" (the base in question was aligned with the reference base as a 'M' CIGAR operation), and the one base in question and the reference base after each included splice boundary were aligned "diagonally".

[00384] To reinforce these restrictions, the combined reference and subject sequences are divided into zones, which are assigned identifiers or zone IDs, for example, integer values. In one embodiment, one is assigned to each dummy base between exon segments, another zone ID is assigned to the last base of each exon segment preceding the splice boundary (but not the last base of the merged sequence), and another zone ID is assigned to all remaining bases of each exon segment.

[00385] For example, for a scaffold with three exon segments (seed chains), each 20 bases long, there could be 4 zone IDs: zone 1 for bases 1-19 of the first exon segment, zone 2 for base 20 of the first exon segment, zone 3 for a false base between the first and second exon segments, zone 4 for bases 1-19 of the second exon segment, zone 5 for base 20 of the second exon segment, zone 6 for a false base between the second and third exon segments, and zone 7 for bases 1-20 of the third segment exon. The same zone mapping is applied to the combined subject sequence and to the combined reference sequence, recognizing that the corresponding subject zones and multi-base reference zones with the same zone ID may have different lengths due to indels in the seed strands. Then, in the intermediate-free aligner, the evaluation cell is modified to allow only valid alignment evaluations at the intersection between identical zone IDs, i.e. when the considered zone ID matches the reference zone ID.

[00386] Accordingly, in various examples, the present invention is directed to devices and methods for use in mapping and aligning DNA and/or RNA. Thus, in specific examples, a hard-wired digital logic circuit, such as an integrated circuit, is provided, wherein the integrated circuit includes a configuration, such as a hard-wired and/or pre-configured configuration, that is configured to implement one or more steps of a mapping and/or alignment operation DNA and/or RNA. In particular, devices disclosed herein may be configured to implement various types of RNA assays, for example, types of RNA assays implemented by one or more hardwired processing modules or a subset thereof.

[00387] For example, in some embodiments, an apparatus and/or system is provided for implementing a DNA and/or RNA sequence analysis pipeline on DNA and/or RNA sequence data, for example, on a read of RNA-derived genomic data. In such an example, the system may comprise one or more of: a memory, for example, for storing one or more DNA and/or RNA reference sequences, for example, RNA-derived genomic reference data, an index of one or more DNU and/or RNA reference sequences, and a plurality of genomic data reads, for example, when each of the reference DNA and/or RNA sequences and the plurality of sequence data reads contain a nucleotide sequence; as well as the integrated circuit disclosed herein. In particular, an integrated circuit may be formed from a plurality of hard-wired digital logic circuits that may be connected to each other via a plurality of physical electrical interconnections. In such an example, one or more of the plurality of physical electrical interconnections may comprise a memory interface for the integrated circuit to access memory. In addition, the hard-wired digital logic circuits may be implemented as a set of processing modules, such that one or more of the processing modules are formed by a subset of the hard-wired digital logic circuits and are configured to implement at least one step in a genomic DNA sequence analysis pipeline and/or or RNA on multiple reads of sequenced data. It should be noted that in various examples, a read of RNA-derived genomic data may refer to a read that was obtained directly by sequencing the RNA sample or by sequencing some further product derived from the RNA sample, such as reverse transcribed cDNA and the like, and may be referred to as herein as an “RNA read” or “long RNA read”, which provides general information about the source from which the RNA data was obtained.

[00388] In particular, the plurality of processing modules may comprise a DNA and/or RNA mapping module, an alignment module, a sorting module, and/or a variant detection module, which may include a Hidden Markov Model (HMM) module and/or a Smith-Waterman (SW) module ). For example, in a first configuration, the hard-wired digital logic circuit disclosed herein may be configured to access in memory, via a memory interface, at least some of the DHE and/or RNA nucleotide sequences in a selected read from a plurality of reads and to an index of one or more DNA and/or RNA reference sequences, and to map the selected DNA and/or RNA read onto one or more segments of said one or more genetic reference sequences based on the index to obtain a mapped read. In specific cases, such as with respect to RNA mapping, the RNA mapping module may be configured to implement one or more of tethered primer mapping, annotated splice boundary searching, and/or primer strand scaffolding and/or the like with respect to stages of RNA mapping.

[00389] Likewise, in a second configuration of hard-wired digital logic circuits, an alignment module may be provided configured to access one or more DNA and/or RNA reference sequences from memory via a memory interface for aligning mapped DNA and/or RNA reads, for example, from a mapping module, to one or more positions in one or more segments of one or more DNA and/or RNA reference sequences to obtain an aligned read. In specific cases, such as with respect to RNA alignment, the RNA alignment module may be configured to implement one or more of splice boundary stitching and/or read-to-splicing alignment, and/or the like with respect to RNA alignment steps.

[00390] Accordingly, in various examples, a hard-wired digital logic circuit may be provided, wherein the digital logic circuit, or a subset thereof, comprises a mapping and/or alignment module, which may be configured to contain a plurality of configured, e.g., preconfigured, modules processing to implement one or more steps in an RNA analysis pipeline, for example, such that one or more steps may include mapping using tethered short primers, searching for an annotated splice boundary, scaffolding primer strands, stitching splice boundaries, aligning a read to a splice, and/or one or more other related steps in implementing mapping and/or alignment operations, such as in a genetic analysis pipeline.

[00391] Additionally, in some examples, a base detection (VC) module is provided, configured, for example, in a third configuration of hard-wired digital logic circuits, with the ability to access an aligned DNA and/or RNA read and at least one of the reference sequences and performing one or more of the following steps. For example, the base detection module may be configured to compare a nucleotide sequence in the aligned DNA and/or RNA read with a nucleotide sequence of at least one genetic reference sequence so as to determine one or more differences between the nucleotide sequence in the aligned DNA/RNA read and the sequence DNA/RNA nucleotides in at least one genetic reference sequence, and with the ability to generate one or more variant definitions representing the one or more differences. Moreover, with respect to the disclosed integrated circuit, one or more of the plurality of physical electrical interconnections may also include an output from the integrated circuit for exchanging resultant data from a mapping module and/or an alignment module and/or a variant determination module.

[00392] In particular, an integrated circuit of the present invention may comprise one or more sets of hard-wired digital logic circuits and/or subsets thereof, for example, including first, second, third, or more sets of configured, e.g., pre-configured, hard-wired digital logic circuits that are configured as one or more processing modules to implement one or more distinct steps in a DNA and/or RNA sequence analysis pipeline. For example, a hard-wired digital logic circuit may comprise a first subset of digital logic circuits that are configured as a processing module to receive a read of DNA and/or RNA data through one or more physical electrical interconnects. In addition, a second subset of hard-wired digital logic circuits may be provided that is configured as a processing module for extracting a portion of a DNA or RNA read to form a seed, for example where the seed represents a subset of the DNA or RNA nucleotide sequence represented by the read, for example, to implement one or more or mapping using tethered short seeds. One or more additional subsets of digital logic may be included, such as a processing module for searching for an annotated splice boundary and/or for performing seed strand mapping. Additionally, a subset of digital logic circuits may be included configured as a processing module, for example, to perform one or more alignment functions on RNA data, including a subset of digital logic circuitry to perform the operation of stitching splice boundaries and/or aligning the read to the splice.

[00393] Accordingly, the integrated circuit of the present invention may include one or more digital logic circuits or a subset thereof for performing one or more steps in alignment using tied short seeds. As discussed in more detail herein, short seed mapping can be performed to improve database sensitivity, e.g., hash table based seed mapping, e.g., by using longer seed matches as anchor points to guide localized searches with more short seeds. This is useful for making hash table-based mapping work well for RNA reads, but it is also useful for improving sensitivity for DNA mapping. In particular, seed mapping may use one or more K-base primers derived from a DNA and/or RNA genomic data read to access a first index, e.g., an index based on a hash table, a DNA and/or RNA reference genome. In such an example, subsequent mapping using anchored short seeds could be performed, for example, using L-base (L<K) seeds to access a second index based on the hash table of the plurality of reference cells, for example, where each of the plurality of reference cells could be a complete or incomplete subset of the reference genome. This may also be useful when using a K and/or L base seed to separately access a first and/or second index, such as a hash table-based index, thereby identifying each one or more linked cells , selected from a variety of reference cells.

[00394] In such an example, one or more data structures may be provided, such as a single or multiple data structure, wherein the data structure(s) may include a first index, such as a hash table-based index, and may also include a second index, for example, a second index based on a hash table. In addition, when one or more anchored cells are provided, identifying the one or more anchored cells may be included as a step in the methods. Therefore, such detection may involve including the identifier of each linked cell in a hash key, for example, a hash key used to access the first and/or second indexes based on the hash table. In some examples, one or more anchored cells, for example, for L base primer hits, can be selected, for example, in a mapping module, based on matches detected by K base primer hits. Accordingly, the mapping module may be configured to implement a mapping operation using anchored short seeds before or after outputting match locations, for example, when one or more anchored cells for L-base seed hits are selected in or outside of the mapping module based on outputting match locations via the mapping module. In some cases, the mapping module may perform mapping using anchored short seeds, where the secondary mapping procedure may traverse at least a subset of the input reads.

[00395] In addition, with respect to performing the alignment, one or more anchored cells for K-base or L-base seed calls may be selected within or outside the mapping module and may be based on the alignment output of the alignment module, or in software software, or in hardware, for example when the alignment module receives match locations from the mapping module. In such an example, a subset of the input reads could be selected, with reads with sufficiently trimmed alignments included or excluded. In some examples, a subset of the input reads may be selected, with reads with alignments with a sufficiently low score included or excluded. And in various other examples, the input reads may be paired-end reads, and a subset of the input reads may be selected to include or exclude pairs of reads with missing alignments in appropriately paired configurations.

[00396] In addition, a mapping module for performing RNA sequence analysis may include one or more processing modules that are configured to search for one or more splice boundaries. Finding splice boundaries can be used to improve the sensitivity of mapping RNA-specific reads. For example, after mapping portions of an RNA read using an RNA reference index, for example, which may or may not be a hash table-based index, a hash table-based index can be generated and queried "database", or any other suitable form of structure data, for example, in the same memory as the reference and/or its index. In particular, a “database” can be generated based on known and/or defined RNA splicing boundaries for the species under study and can be accessed based on mapped positions. It should be noted that each known splice boundary may represent a possibly long "intron" segment (up to 1Mbp or longer) in the reference, for example, where read alignments will typically "jump" from one intron endpoint to the other.

[00397] Accordingly, the database may be accessed in this manner to retrieve known splice boundaries that have one endpoint at or near each of the reference segments along already mapped portions of the read, but which may have another portion of the read that traverses beyond the adjacent intronic endpoint. In such examples, it can be tentatively assumed that the specified other portion of the read will then continue to match the reference after jumping through the intron, even though previous mapping attempts may have failed to detect any match in the specified region. Later splice alignment and scoring can then measure how well the read actually matches the alignment at the specified splice boundary. This method can thus improve the ability of the integrated circuit to detect the most likely spliced read alignment despite obstacles such as short exons, short exon overhangs and/or edits (SNPs, etc.) blocking a match in an exon segment middle length.

[00398] Accordingly, a memory configured to be coupled to an integrated circuit may be provided, the memory comprising an index of a reference genome and a list of annotated splice boundaries within the reference genome. The mapping module may thus map the first portion of the RNA-derived genomic data read to a match location in a reference genome, for example, by accessing the reference index using the first portion of the read. The mapping module may then access the list of annotated splice boundaries and retrieve the intron descriptor, for example when the intron descriptor includes a first endpoint and a second endpoint in the reference genome, for example when the first endpoint is within a limited distance of the match location in the reference genome .

[00399] The mapping module may then map the third portion of the RNA-derived genomic data read to a putative location in the reference genome, for example, where the putative location in the reference genome may be adjacent to a second endpoint in the reference genome of the intron descriptor. The mapper then outputs mapped read locations of the RNA-derived genomic data, for example, where the mapped RNA-derived genomic data includes at least a location of a match in a reference genome and a predicted location in the reference genome. Accordingly, in various examples, a list of annotated splice boundaries may be provided, which list may be formulated as a table containing an entry for one or more, for example, each of a plurality of reference bins, thereby forming a partition of the reference genome. The list of annotated splice boundaries may then be accessed in a manner that includes identifying at least one reference bin within a limited distance of the match location in the reference genome and accessing table entries that correspond to the at least one reference bin.

[00400] In such an example, the actual location of a splice boundary in a read can be determined, for example, by using a first portion of the read, eg, matching a location in a reference genome, by using a first endpoint and a second endpoint of an intron descriptor in the reference genome. Additionally, in some examples, the limited distance may be defined to at least ensure that the actual location of the splice boundary in the read is not outside the boundaries of the read. Accordingly, a first strand of primers can be determined, for example using at least a first portion of the RNA-derived genomic data read and the location of the match in a reference genome. A second strand of primers may then be determined, for example, by using at least a second portion of the RNA-derived genomic data read and a predicted location in a reference genome. A relationship can then be established between the first seed strand and the second seed strand, and the output data can be mapped to RNA-derived genomic data read locations that include descriptions of the first seed strand, the second seed strand, and the relationships between them.

[00401] In addition, a mapping module for performing RNA sequence analysis may include one or more processing modules that are configured to perform the operation of scaffolding seed strands. It should be noted, however, that while reference is made herein to seed chains, such seed chains are not limited to the context of hash table-based seed mapping, but the concept can be extended to any mapping of a sub-read to a segment of a reference genome. Such scaffolding operations are useful for translating a list of seed chains into a list of scaffolds, for example, when each scaffold is a sequence of one or more seed chains that represents a likely alignment with a spliced read.

[00402] In various examples, successive portions of a read may be mapped to successive segments of a single reference chromosome, for example, in a consistent orientation and order. Accordingly, the method of forming multiple scaffolds can be selective, since essentially for a list of seed chains, the number of combined sequences and the number of seed chains can be very large. Thus, the list of scaffolds can be generated to be large enough to include the correct alignment to the spliced read with a high degree of confidence, for example, without generating too many false scaffolds. In such an example, each scaffold can then be assessed, for example by splicing alignment.

[00403] Accordingly, the mapping module of the present invention may be configured to determine a list of seed strings such that each seed string of the list of seed strings represents a match between a corresponding portion of the RNA-derived genomic data read and a corresponding segment of a reference genome. By examining the list of seed chains, the mapper can then generate a list of scaffolds, for example, wherein each scaffold may contain one or more individual seed chains from the list of seed chains, and/or the scaffold suggests a corresponding sequence portion of the read of one or more corresponding portions of the RNA-derived genomic read data. The sequence of parts of the read can evolve in a single direction in the read of RNA-derived genomic data. Likewise, the scaffold may also assume a corresponding reference segment sequence of one or more corresponding reference genome segments, such that the evolution of the reference segment sequence occurs in a single direction in the reference genome.

[00404] In addition, obtaining a list of scaffolds may further include sorting the list of seed chains in ascending or descending order of corresponding reference genome segments, for example, where the list of scaffolds is obtained in accordance with a set of rules that determines when one seed chain can follow another chain seeds in the scaffold. Such a set of rules includes the minimum and/or maximum allowable gap and/or minimum and/or maximum allowable overlap between successive read parts in a sequence of read parts. In some examples, the sequence of the read portions and the sequence of the reference segments may imply that the intron length sequence of the calculated alignment diagonals is shifted from each portion of the read and the corresponding reference segment to the next portion of the read. In addition, the set of rules may include a minimum allowed intron length and a maximum allowed intron length for an intron length sequence.

[00405] Additionally, in various examples, executing a scaffold list may involve executing an initial part of the scaffold having a partial sequence of one or more individual seed chains, and may further include executing a sequence of two individual scaffolds in the scaffold list such that one or more of the at least two separate scaffolds may represent extensions of the initial part of the scaffold to longer scaffolds. Such a list of scaffolds may include filtering out smaller scaffolds, e.g., as being smaller according to a calculated filter score, e.g., a filter score calculated using the difference between each scaffold's RNA-derived genomic read coverage and/or the maximum RNA-derived genomic read coverage data, for example, maximum coverage calculated over the entire list of scaffolds. In some examples, the filtering score can be calculated using the coverage of each scaffold in the reference genome.

[00406] As discussed above, in various examples, a hard-wired digital logic circuit may be provided, wherein the digital logic circuit, or a subset of the logic circuits, includes an alignment module that may be configured to contain a plurality of processing modules for implementing one or more alignment steps in an RNA analysis pipeline, for example, such that one or more steps may include stitching splice boundaries and/or aligning the read to the splice. In particular, an alignment module for performing RNA sequence analysis may include one or more processing modules that are configured to perform one or more splice boundary stitching operations and/or one or more splice read alignments.

[00407] Specifically, in various examples, a pair of partial mappings, e.g., primer strands, e.g., two consecutive primer strands in a scaffold, may be generated and/or otherwise provided in an RNA read. In various examples, partially mapped primer strings may represent two exon segments of a likely spliced mapping that skips a possible intron in the reference. In such an example, the alignment module processing module may be configured to implement a splice boundary stitching operation that is designed to accurately, for example accurately, more or less accurately determine the most likely position in the read where an intron has been skipped. The result is the stitching position between the two bases of the read such that bases to the left of the stitching point align with the first exon segment, and bases to the right of the stitching point align with the second exon segment. This can be accomplished by 1) testing a variety of possible stitching positions and 2) evaluating the test results. Such an assessment may be based on a variety of different criteria, for example, the number of base mismatches observed and/or the absence or presence and type of canonical intron motifs observed at the two ends of the putative intron span in the reference, for example, according to a given cross-linking position.

[00408] The splice stitching operation can be configured as a preprocessing procedure for spliced alignments that would otherwise be resource-intensive and/or expensive to perform if the operation needs to be performed in a manner that considers all stitching positions. Accordingly, the alignment module may receive splicing mapping for an RNA-derived genomic data read, wherein the splicing mapping includes at least a first portion and a second portion of the RNA-derived genomic data read and at least a first segment and a second segment of a reference genome.

[00409] In addition, the alignment module may be configured to implement a splice stitching operation so as to determine the best splice position within a read of RNA-derived genomic data. Such best stitching position can be realized by optimizing a plurality of stitching factors that relate to each stitching position considered. Crosslinking factors may include the degree of match between the first part of the read, for example, matched in length to the end of the stitching position in question, and the first segment of the reference genome, matched in length identically. Cross-linking factors may also include the degree to which the second portion of the length-matched read matches the start of the crosslink position in question, and the second segment of the reference genome with an identical length match. In this case, the cross-linking factors may also contain the probability of an intron motif corresponding to the cross-linking position in question, for example such that the intron motif includes at least two reference bases adjacent to the length-matched first reference segment, and at least two reference bases adjacent to second reference segment consistent in length.

[00410] In various examples, the stitching factors may be combined into a score and the numerically best stitching position in question may be determined to be the best stitching position. In some examples, a transition may be made from a first crosslinking position of interest to a second crosslinking position of interest via at least one interfering nucleotide in a read of RNA-derived genomic data. In such an example, a score for the second cross-link position under consideration may be calculated in part by adjusting the score for the first cross-link position under consideration to account for any difference between how well the at least one interfering nucleotide matches the first segment of the reference genome and how well the at least one interfering nucleotide matches the nucleotide interference coincides with the second segment of the reference genome. In some examples, the best stitch position may be passed to a stepless alignment module, which may be configured to determine the best stepless alignment of a read of RNA-derived genomic data combining the RNA-derived two length-matched reference genome segments. The best stitch position can then be passed to the gap alignment module, which can then determine the best gapless or gap alignment of the RNA-derived genomic reads, combining the RNA-derived two length-matched reference genome segments.

[00411] The alignment module may also include a module configured to implement spliced read alignment. For example, when performing an alignment of a spliced read, a sequence of multiple partial mappings, eg, strands of primers, eg, sequential strands of primers in a scaffold, can be generated in the RNA read. A plurality of partial maps may represent multiple exon segments of candidate splice maps that may miss one or more candidate introns in the reference sequence (eg, undergoing splice boundary stitching for each intron). In such examples, the splicing alignment operation can be performed by aligning the read(s) to concatenate multiple exon segments in the reference.

[00412] In particular, an alignment module may be provided and configured to receive splicing mapping for one or more reads of RNA-derived genomic data. Splicing mapping may involve determining the sequence of multiple adjacent portions between RNA-derived genomic data reads and one or more corresponding sequences having one or more, eg, multiple, reference genome segments. The alignment module can then implement a splice alignment operation on the specified sequences to determine the best splice alignment of the RNA-derived genomic data read to the sequence of multiple reference genome segments. For example, a splicing alignment operation may comprise combining multiple reference genome segments into a combined reference sequence in which each of the multiple reference genome segments is joined at the join junctions. In addition, a read sequence having at least a plurality of adjacent read portions derived from the RNA genomic data can be generated and joined at such join junctions.

[00413] The best sequence alignment of a read can be calculated with respect to a combined reference sequence, for example, when the best sequence alignment is constrained such that the join junctions in the read sequence are aligned with the corresponding join junctions in the merged reference sequence. The sequence alignment may be edited into a spliced alignment, which editing may include insertion of intron descriptors at positions in the alignment corresponding to junction junctions, for example such that the intron descriptors encode intron length, which may be equal to the corresponding distances between reference genome segments. The alignment module can then output the spliced alignment.

[00414] In such an example, the best sequence alignment may be defined as the alignment with the numerically best score among the computed scores for all likely alignments, the score of each likely alignment being calculated to include mismatch penalties for each nucleotide of the read sequence that did not match with the aligned nucleotide of the combined reference sequence. The likely alignment score may also be calculated to include one or more other penalties, such as an indel penalty for each insertion or deletion in the likely alignment and/or a splice penalty for each join junction included in the likely alignment, such that the splicing penalty is determined at least in part according to the intronic motif associated with joining the reference genome segments. In various examples, the merged reference sequence may be configured to include false bases, for example, at fusion junctions, where the false bases may be used to carry the values of the corresponding splicing penalties.

[00415] In these examples, computing the best sequence alignment may involve dynamic programming to compute cell scores for a two-dimensional matrix of score cells, said two dimensions corresponding to the read sequence and the combined reference sequence. In such examples, each of the multiple adjacent parts of the read may be assigned a zone ID; and, in addition, each of the plurality of reference genome segments may be assigned a zone ID. Such a zone ID may be configured to be similar to each corresponding portion of the read and the reference segment, for example such that when each evaluation cell in a matrix of evaluation cells has a cell read zone ID that is similar to the zone ID of the corresponding portion of the read. In addition, the cell's reference zone ID can be configured to be similar to the zone ID of the corresponding reference segment, and the best sequence alignment can be constrained to only pass through evaluation cells whose cell read zone IDs are similar to their cell's reference zone ID.

[00416] In FIG. Figure 5 shows an abstract alignment rectangle in which the merged reference sequence runs along the vertical axis and the merged reference sequence runs along the horizontal axis. The false bases of each merged sequence are shaded (zones 3 and 6). A grid is superimposed on the alignment rectangle to show the boundaries between zones on each axis. The auxiliary rectangles with matching zone IDs are valid alignment areas, and the other (shaded) auxiliary rectangles are invalid alignment areas. Shown is an example of a valid alignment that goes end to end in the sequence in question and contains an insertion (vertical segment) in the second exon segment (zone 4) and a deletion (horizontal segment) in the third exon segment (zone 7). The actual alignment runs diagonally across the splicing boundaries (zones 3 and 6).

[00417] Intermediate or gap alignment using a combined target and reference sequence produces correct alignment estimates, but the alignment trace (eg, CIGAR line) requires editing because it does not yet include intronic ('N') operations. For example, a spliced alignment of a 100-base read without indels could come from a “101M” CIGAR alignment, meaning that 101 bases are aligned diagonally without indels. The above CIGAR requires two adjustments. First, false bases between exon segments are counted in CIGAR, but they shouldn't be. Second, an intronic operation, for example 895 bases long, must be inserted at a false base position. A valid CIGAR could be, for example, "40M895N60M".

[00418] Assuming that the scaffold containing the seed chains defining the endpoints of the exon segments is stitched, then through straightforward arithmetic one can discover the position of each splice boundary in CIGAR, remove the "1M" for the false base, and replace it with an intronic operation of the proper length. For local alignments, this process may account for the possibility that one or more of all exon segments have been excised from the alignment. The same arithmetic process can calculate the correct start and end positions of the alignment in the reference genome.

[00419] Once alignment scores, start and end positions, and CIGAR strings are obtained for each aligned scaffold, the processing to select and return the best alignment, possibly with splicing, is identical to DNA processing. Pairwise end candidate alignments are examined to find properly positioned and oriented alignment pairs. Probable alignment pairs, including inappropriately formed probable pairs, are given score penalties for being unpaired or having unlikely empirical insertion lengths; pairing scores are formed by combining (eg, adding) the alignment scores of each mate and the pairing penalty; and the pair of alignments with the best score is selected and provided output from the alignment module.

[00420] The apparent length of an insert, typically measured as the coverage in the reference covered by the alignments of two mated reads, may be very long due to introns in one or both of the mated reads or unobserved introns in the gap between the mated reads. (Physical insertions are potentially much shorter and have the length of a sequenced RNA or cDNA molecule in which the introns have been removed by splicing.) Thus, significantly longer apparent insertion lengths should be considered to pair properly and should be assigned zero or small formation penalties steam; this can be done according to the distribution of known intron lengths in the species from which the sample is derived and/or the distribution of observed apparent indels in the RNA-seq data being processed.

[00421] In one embodiment, for each read processed, an alignment estimate, a starting position, and a CIGAR encoded string are provided from the alignment module. Additionally, in another embodiment, the intron motif and annotation status are provided for each splice boundary in the alignment. The quality or reliability of the mapping, for example the "MAPQ" parameter on the phred scale, can also be assessed and reported. In a preferred embodiment, the MAPQ is pre-scored as a proportion of the difference between the best pair estimate and the second best pair estimate with a different alignment for the current read. Additional candidate alignments, or secondary alignments, may also be returned for each read, for example, a limited number of other candidate alignments whose score is within a defined and configured score difference threshold.

[00422] It should be understood, with respect to the above, that while in some cases a mapping function has been described, for example as a mapper, and/or in some cases an alignment function has been described, for example as an aligner, these various functions may be performed sequentially by one and the same the same architecture, which in the art is usually called an equalizer. Accordingly, in various cases, the mapping function and the alignment function as described herein may be performed by a common architecture, which can be understood as an aligner, especially in cases where in order to perform the alignment function, it is necessary to perform the mapping function first.

[00423] The output from the alignment module is a SAM (text) or BAM file (e.g., a binary version of a SAM file) along with a Mapping Quality Score (MAPQ), which reflects the confidence that the predicted and aligned location of the read relative to the reference is actually the same one. where did the read come from? Accordingly, once it is determined where each read is mapped, and it is also determined where each read is aligned, for example, each corresponding read is given a position and a quality score reflecting the likelihood that that position is a correct alignment, so that the nucleotide sequence for the DNA a subject's DNA is known, as well as how the subject's DNA differs from a reference (eg, a CIGAR string is defined), the various reads representing the subject's genomic nucleic acid sequence can be sorted by chromosomal location so that the precise chromosomal location of the read can be determined. Therefore, in accordance with some aspects, the present invention relates to a sorting function, for example, which can be performed by a sorting module, which can be part of a pipeline of modules, for example, a pipeline designed to receive raw sequence read data, for example, in the form of a genomic sample of an individual, and mapping and/or alignment of this data, which can then be sorted.

[00424] In particular, after reads are assigned positions, for example, relative to a reference genome, which may include determining which chromosome the read belongs to and/or its offset from the start of that chromosome, the reads can be sorted by position. Sorting can be useful, for example, in subsequent analyzes, since with its help all reads that overlap a given position in the genome can be formed into a cluster so that they are located next to each other, for example, after processing by the sorting module, as a result of which it is possible It is easy to determine whether the majority of reads agree with the reference value or not. Thus, if the majority of reads do not agree with the reference value, the variant definition can be flagged. Therefore, sorting may involve one or more sortings of reads that are aligned to the same position, e.g., the same chromosome position, to create a cluster so that all reads that cover the same location are physically grouped together; and may also include analysis of reads in the assemblage to determine where the reads may indicate an actual variant in the genome compared to a reference genome, which variant can be distinguished, for example, by assemblage reconciliation, from an error such as a machine read error or an error in sequencing methods, which may manifest itself in a small part of the reads.

[00425] Once the data is received, there are one or more other modules that can be used to clean up the data. For example, one module that may be included, for example, in a sequence analysis pipeline, such as to determine the genomic sequence of an individual, may be a local realignment module. For example, it is often difficult to identify insertions and deletions that occur at the end of a read. The reason is that the Smith-Waterman algorithm or similar alignment process lacks context beyond the indel to allow evaluation to detect its presence. Therefore, the actual indel may be reported as one or more SNPs. In such a case, the accuracy of the predicted location for any given read can be improved by performing local realignment on the mapped and/or aligned and/or sorted read data.

[00426] In such cases, clusters can be used to help figure out the proper alignment, for example, when a position in question is at the end of any given read, that same position is likely to be in the middle of some other read in a given cluster. Accordingly, when performing local realignment, various reads in a cluster can be analyzed to determine whether some reads in the cluster indicate the presence of an insertion or deletion at a given position where another read does not contain an indel or rather has a substitution at that position, then one can insert an indel, for example, into a reference where it is not present, and one can re-align reads in a local cluster that overlaps that region to see if this achieves a better aggregate score than if it were not there insertions and/or deletions. Accordingly, if there is an improvement, the entire set of reads in the cluster can be revised, and if the assessment of the entire set has improved, then it becomes clear that there was indeed an indel at this position. In a similar way, one can compensate for the lack of sufficient context to more accurately align the read at the end of the chromosome for any individual read. Therefore, when performing a local realignment, one or more clusters where one or more indels may be present are examined and it is determined whether the overall alignment score can be improved by adding an indel at any given position.

[00427] Another module that may be included, for example, in a sequence analysis pipeline, such as to determine the genomic sequence of an individual, may be a duplicate tagging module. For example, a duplicate flagging feature can be implemented to compensate for chemistry errors that may occur during the sequencing phase. For example, as described above, during some sequencing procedures, nucleic acid sequences are attached to beads and extended onto them using labeled nucleotide bases. Ideally, you will get one read per bead. However, sometimes multiple reads are attached to a single bead, resulting in an excess number of copies of the attached reads. This phenomenon is known as read duplication.

[00428] Such duplication of reads can confuse statistics and create statistical bias because, rather than having an equivalent representation of all reads, different reads have been duplicated, for example due to duplicate template sequences attached to more than one bead being represented in excess. Accordingly, this can be determined since any read that aligns to the exact same position and has exactly the same length is most likely a duplicate. Once this is determined by the system, only one read needs to be further processed and others may be marked as duplicates and therefore discarded or ignored. A typical situation in which this occurs is when there is not enough genetic material to process to begin with and the system attempts to overcompensate.

[00429] Another module that may be included, for example, in a sequence analysis pipeline, such as for determining the genomic sequence of an individual, may be a base quality estimate recalibrator. For example, each base of each read has a Phred score, which indicates the probability that a particular base at a specified position is incorrect. For example, the Phred score for any base is due in part to the nature of the base that precedes it, and the error profile will differ depending on which base precedes the base in question. In addition, there is a high probability of an error occurring at the ends of the read, for example, when the chemistry begins to lose its qualities at the ends of the read. Recalibration of the foundation quality estimate is a covariate of analysis that can go back and measure the empirical qualities of the foundation quality estimate as a function of all the things that change it.

[00430] In various examples, this involves two passes, the first collecting all the actual, empirically measured data and observed error rate statistics as a function of all variables, and the second pass involving the actual recalibration of the scores by running all reads through a filter that modifies the quality scores for each individual grounds as a function of variables based on what was actually measured empirically in the data set. This compensates for any differences in data due to different variables and cleans up all data and estimates. The purpose of all of these purifications is to ensure that the best possible definition of variants is achieved. Many variant detectors base their decisions in part on the reported quality of each nucleotide that accumulates at each position in the genome. If quality assessments are not accurate, the result can easily be determined to be the wrong basis.

[00431] Another module that may be included, for example, in a sequence analysis pipeline, such as to determine the genomic sequence of an individual, may be a compression module that performs a compression function. As stated above, it may be useful at some stage to take the generated and processed data and transmit it to a remote location, such as the cloud, and therefore the data may need to be compressed at some stage of processing so that once it is compressed, it can be transmitted and/or uploaded in another way, for example, to the cloud or to a group of servers, etc., for example, to implement the base determination module. The obtained results can then be decompressed and/or stored in memory, in a database in the cloud, for example, an electronic health database and/or a research database and the like, which in turn can be made available for tertiary processing, etc.

[00432] In particular, after genetic data has been generated and/or processed, for example, in one or more primary and/or secondary processing protocols, for example, mapped, aligned and/or sorted, for example, to create one or more files determining variants, for example, to determine how a subject's genetic sequence data differs from one or more reference sequences; in another aspect, the present invention may relate to performing one or more other analytical functions on generated and/or processed genetic data, for example, to further processing such as tertiary processing. For example, the system may be configured to perform further processing on the generated and/or secondarily processed data, such as by running it through one or more tertiary processing pipelines, such as one or more of a genome pipeline, an epigenome pipeline, a metagenome pipeline, a shared genotyping, MuTect2 pipeline or other tertiary processing pipeline, for example, through the devices and methods disclosed herein. For example, in various examples, an additional level of processing may be provided, for example, for disease diagnosis, therapeutic treatment and/or prophylactic prevention, including, for example, non-invasive prenatal testing (NIPT), neonatal intensive care unit (NICU), cancer, performed in the laboratory research (LDT), agrobiology (AgBio) and other types of diagnosis, prevention and/or treatment of such diseases that use data generated by one or more of the specified primary, secondary and/or tertiary pipelines. Therefore, the devices and methods described herein can be used to generate genetic sequence data, which can then be used to generate one or more variant definition files and/or other associated information that can be further processed by other tertiary pipelines. processing in accordance with the devices and methods described herein, for example, for the diagnosis of specific and/or general diseases, as well as for preventive and/or therapeutic measures and/or developmental methods.

[00433] Accordingly, as discussed hereinabove, in various aspects, the present invention relates to systems, devices, and methods for implementing genomic and/or bioinformatics protocols, such as, in various examples, for performing one or more functions for analyzing genetic data on an integrated circuit, for example, implemented on a hardware processing platform. For example, in one aspect, a bioinformatics system is provided that may include implementations of various bioinformatics functions that have been optimized for faster implementation and/or for implementation with increased accuracy in hardware. Accordingly, in various examples, the methods and systems disclosed herein may include implementing one or more algorithms to perform said functions, which algorithms may be implemented as a hardware solution, for example, such that the algorithms are optimized to perform by means of an integrated circuit formed from one or more hard-wired digital logic circuits. In such an example, hard-wired digital logic circuits may be interconnected, for example, through one or a plurality of physical electrical interconnections, and may be configured to function as one or more processing modules. In various examples, a plurality of hard-wired digital logic circuits are provided that are configured as a set of processing modules, each processing module configured to implement one or more steps in a bioinformatics genetic analysis protocol.

[00434] In particular, according to one aspect, a system is provided for implementing a sequence analysis pipeline, for example, on genetic sequence data. The system may include one or more electronic data sources, a memory, and an integrated circuit. For example, in one embodiment, there is an electronic data source that may be configured to provide one or more digital signals, for example, digital signals representing one or more reads of genetic data, for example, where each read of genomic data includes a nucleotide sequence. In addition, the memory may be configured to store one or more genetic reference sequences and may also be configured to store an index, such as an index of the one or more genetic reference sequences.

[00435] Also in various examples, one or more of the plurality of physical electrical interconnects may comprise an input to, for example, an integrated circuit, and may also be coupled to an electronic data source to enable reception of one or more genomic data reads. In various embodiments, the hard-wired digital logic circuits may be implemented as a set of processing modules, such that each processing module may be formed by a subset of the hard-wired digital logic circuits and may be configured to implement one or more stages in a sequence analysis pipeline, e.g. on digitized genetic data, for example on multiple reads of genomic data. In such examples, each subset of hardwired digital logic circuits may be configured to be wired so as to perform one or more steps in a sequence analysis pipeline, for example, where one or more steps may include performing one or more of a base detection operation and/or correcting an error, for example, in digitized genetic data, and/or may involve one or more of performing mapping, alignment, and/or sorting functions on the genetic data. In certain cases, the pipeline may include performing one or more of realignment, deduplication, base quality assessment recalibration, reduction, and/or compression and/or decompression on the digitized genetic data. In certain cases, the pipeline may include performing a variant detection operation on genetic data.

[00436] Thus, various implementations, systems, apparatus and methods for implementing genomics and/or bioinformatics protocols discussed herein may involve using processes that can typically be performed in software and incorporating said functionality into an integrated circuit, such as a chip 100, for example, as an integral part of a motherboard 105, such that such functions are optimized to improve its on-chip performance. In this regard, in one embodiment, as can be seen in FIG. 6 and 7, a chip 100 is provided, wherein the chip 100 is designed to efficiently perform these software pipeline functions. In various specific embodiments, the chip 100 may be a field programmable gate array (FGPA), an application-specific integrated circuit (ASIC), or a special-purpose integrated circuit (sASIC), or the like.

[00437] Thus, for example, the execution of one or more algorithms may be built into a chip, such as an FGPA or ASIC or fabricated ASIC chip, and may be optimized to operate more efficiently through implementation in hardware. Accordingly, in one embodiment, an FGPA chip is provided, wherein the chip may be configurable, e.g., the software portion of the chip may be modified to be more adaptable to the specific needs of a user to perform various genomic functions described in detail herein . In this case, the user can change or modify the algorithms involved depending on the key parameters designated for the system as a whole, for example, to provide additional functionality or replace what was previously on the chip, for example, to reconfigure the chip in order to enable different algorithm.

[00438] Further, in another embodiment, an FGPA or fabricated ASIC is provided, wherein the chip is fully or partially configurable, for example, some of its software can be changed to make it more adaptable to a user's specific needs. in terms of performing various genomic functions described in detail herein. According to another embodiment, an ASIC is provided, for example converted to an FGPA or sASIC, the functionality of which may be locked into the chip. In such a case, various parameters, such as various parameters related to a particular function of one or more of the algorithms set forth herein, may be selectable by the user, such as controlling how the various modules should function, and wherein the specific mechanism for the operation of such modules is blocked.

[00439] In various embodiments, as shown in FIG. 6 and 7, chip 100 may be part of a system board, such as part of an expansion board 104, such as a peripheral component interconnect (PCI) card, including a PCIe card, which in various embodiments may be integrated, such as coupled with signal carrying capability, e.g., electrically coupled to an automated sequencing device so as to function as an integral part of the sequencer such that data files, e.g., FASTQ format files generated by the sequencer are transferred directly to the chip for secondary genome processing as immediately following file generation in FASTQ format and/or primary processing, for example, immediately after the sequencing function is performed.

[00440] Thus, in some cases, a PCI card 104 is provided, where such a PCI card may include a PCIe chip 105, wherein the card 102 and/or chip 100 may include one or more configuration management modules, such as configuration controller (Cent-Com); direct memory access device (for example, driver); application program interface (API); client level interface (CLI); library; a memory unit, such as random access memory (RAM) or dynamic random access memory (DRAM); and/or chip-level interconnect such as DDR3. Thus, for example, in various cases, a configuration management module may be included, where such a configuration management module is controlled, for example, by a parameters file. In such a case, the configuration management module can be adapted to adjust the configuration of various software pipeline modules. In various cases it may be editable, allowing the user to define which modules of the software pipeline will be used, for example all or a subset of the modules in a smaller number than all, for example, for some set of data, such as some set of files in FASTQ format.

[00441] For example, in various embodiments, the operation of the software pipeline is highly configurable, such that one or more modules, such as on-chip modules, can be run or not run as required. Further, each module used can also be configured to run according to one or more pre-selected parameters that can be controlled by the user, for example relating to how the module will execute and function. Thus, there may be two different sets of configuration files, such that, for example, one controls the basic operations of the system as a whole and can be hidden from the user, and the other is controlled by the user and thus allows the selection of different parameters to be triggered by one or more subsystems, such as a module, chip 100, and/or PCI card 104.

[00442] Thus, the various modules described above may be permanently connected to the chip, or external to the chip but interfaced with the chip, for example on a PCI card 104, or remote to the chip. to the chip, for example, on another PCI card or even another server, for example, on a server accessible via cloud 30. For example, in some implementations, one or more of the modules described above may be permanently coupled to chip 100, and the chip itself is mounted on the circuit board 104 of the stand-alone device 300, or interfaced with the sequencer, and the user configures and runs the system himself according to his own preselected parameters. Alternatively, as discussed herein, one or more of the modules described above may reside on a system accessible via cloud 30, wherein controlling the operation of the software pipeline and/or related modules may include logging a user into the server, for example ,on a remote server, and transmitting data back and forth, and,thus selecting which modules to run on the data set. In some cases, one or more modules may be executed remotely, for example through a cloud-accessible server.

[00443] In some cases, when configuring a system, a chip, such as chip 100 on expansion card 104, such as a PCI card, may be included in server 300 such that various system applications are run from that server. In some cases, the server 300 may include a plug-in terminal with a windowed interface provided to the user so that the user can select modules to run and the parameters by which they are launched, for example, by selecting one of the blocks from a menu of blocks. In other cases, however, the parameters file may be a text file detailing categories of modules under file names, which the user can subsequently edit to select which modules will run and according to which parameters. For example, in various embodiments, each chip may contain all of the modules or a selectable set of modules, such as one or more of the following: basefinding, error correction, mapping, alignment, sorting, local realignment, duplicate flagging, recalibration, variant detection, a compression module and/or a decompression module from which the user can select which modules will run, when and to what extent they will run, without changing the functioning of the underlying algorithms by which the individual modules operate.

[00444] Additionally, in various cases, a direct memory access (DMA) device on the chip and a DMA driver may be included, the DMA driver comprising code that runs in the kernel. Thus, the DMA driver can be at the core of an entire operating system. For example, virtual user space may be a level above the literal addressing space in which the kernel runs code. Such operating system software therefore operates between these layers, managing the transformation of data from virtual space to physical space. Specifically, the core is the lowest level of code that provides the platform with access to PCI 104, such as PCIe, bus 105 to which chip 100 interfaces. Thus, since in various embodiments chip 100 may be configured as in expansion card 104 with PCIe expansion bus 105, and such card 104 can be interfaced with various hardware of some device, such as a sequencer, and the DMA driver can function to communicate with the sequencer hardware and can then be configured to run at the kernel level on the CPU 100, so as to also communicate with the DMA device in the chip 100, and/or operate in virtual user space, so as to receive commands from the user.

[00445] In order to enable such messaging on the chip and/or between the chip and one or more boards, each individual configurable module parameter may be assigned to an address in a register. In such a case, the board may have its own address space, which address space may be different from the address space for one or more memory blocks, for example, 64 GB of memory, and/or each module may additionally contain registers and associated local memory - each with its own address space. Thus, the driver knows where everything is, all the addresses, and knows how to communicate between chip 100, PCI card 104, and/or server hardware. Next, knowing the location of all the addresses and communicating with the API, the driver can read the user-created parameter file and look for the parameter where the file is actually located on the host computer system, and will read and interpret the value written in this file and deliver it value into the proper register at the proper location on the chip. Thus, the driver can control the delivery of commands associated with a selected parameter, such as those related to a user-selected configuration, and deliver such data to the chip via a DMA device to configure its operating functions.

[00446] In particular, after the genetic data has been generated and/or processed, for example, using one of the primary and/or secondary protocols, for example, mapped, aligned, and/or sorted, for example, to create one or more files determination of variants, for example, to determine how the genetic sequence data of the subject of interest differs from one or more reference sequences, and a further aspect of the invention may be associated with performing one or more analytical functions on the generated and/or processed genetic data, such as further ,eg tertiary ,processing. For example, the system configuration shown in FIG. 8-11 may be configured to further process the generated data and/or data resulting from secondary processing, for example, to run through one or more tertiary processing software pipelines 700, for example, through one or more genome software pipelines, software pipelines epigenome, metagenome software pipelines, co-genotyping, MuTect2 software pipelines, or other tertiary processing software pipelines, such as the devices and methods described herein. For example, in various cases, an additional level of processing 122 may be provided, for example, for disease diagnosis, therapeutic treatment and/or prophylactic actions, for example, including NIPT, NICU, Cancer, LDT, AgBio and other similar methods for disease diagnosis, prevention, and /or treatments that involve data generated by one or more of such primary and/or secondary and/or tertiary software pipelines.

[00447] Thus, the devices and methods described herein can be used to generate genetic sequence data, which can further be used to create one or more variant definition files and/or other related data, which can then be the subject of research for other tertiary processing software pipelines being launched in accordance with the devices and methods described herein, for example, both for diagnosis of a specific and/or systemic disease and for prevention and/or therapeutic treatments and/or techniques under development.

[00448] Further, in various cases, an API may be included, where such API is configured to contain a list of function calls that a user can make so as to configure and manage the system. For example, an API may be defined in a header file that describes the functionality and specifies the way in which a function is called, such as the parameters passed, input and output data, what goes in, what goes out, and what is returned. For example, in various implementations, one or more elements of the software pipeline may be configurable, such as commands entered by the user and/or one or more third-party applications. Such commands can be sent to the chip via an API, which communicates with the driver, telling the driver which components, such as modules, should be activated, when and in what order, based on pre-selected configuration parameters.

[00449] As stated above, the DMA driver runs at the kernel level and has its own simple very low level API that provides access to hardware and functions in order to access applied registers and modules. Above this layer is a virtual service function layer, which forms the building blocks used for a variety of functions that send files to the kernel and receive results, and subsequently performs higher-level functions. Above this layer is an additional layer that uses these utility functions, which is the API layer that the user interacts with and functions primarily for configuration, file loading, and results uploading. Such configuration may involve exchanging information with registers as well as making function calls.

[00450] For example, as described above herein, one of the function calls may be made to create a placement table using a shuffle algorithm. Moreover, since in some implementations this function can be created based on one control genome, and only once for each control genome, and there may be a need to create arrangement tables used in the mapping module based on the control sequence, then there is a function call that performs such function, and which will receive the file name associated with the location of the control file and will then create one or more data files containing the layout table and control sequence. Another function call may serve to load the arrangement table created by the shuffle algorithm and transfer it to memory on chip 100 and/or place it in an appropriate location determined by the hardware. Of course, it will be necessary to load the check sequence onto the chip 100, which is also required for the alignment function to work. The configuration management module may perform such a function, for example, by loading everything necessary for the modules of the chip 100 to perform their functions in memory on the chip or attached to the chip 100.

[00451] Additionally, the API may be configured to allow chip 100 to communicate with the sequencer motherboard, if included, so as to receive FASTQ format sequence files directly from the sequencer, for example, immediately after they are generated, and then pass this information to the configuration management module, which then forwards this information to suitable data stores in the hardware 100, which make this information available to the corresponding modules in the hardware so that they can perform their assigned functions, such as search for bases, map, align, sort, etc. DNA sample relative to a reference genome.

[00452] Moreover, a client level interface (CLI) may be included, where such a CLI may allow a user to call one or more of the specified functions directly. In various implementations, the CLI may be a software application tailored to configure the use of hardware. Thus, a CLI can be a program that accepts commands, such as arguments, and makes all functionality available through a simple application call. As stated above, CLI can be implemented based on the command line or GUI (graphical user interface). Line commands can operate at a level below the GUI, while the GUI includes a windowed file manager with the function of selecting blocks with the mouse, which schematically depict the modules used and the parameters for their use. For example, if the appropriate commands are received, the CLI will, in its operation, determine the location of the control sequence, determine whether there is a need to create a lookup table and/or index, and, if they have already been created, determine where to store them, and direct the unloading of the created lookup tables and /or pointer, etc. Commands of this type can appear at the user's discretion when working with the GUI, if the user selects the described chip to work with.

[00453] In addition, a library may be included, where such a library may contain previous editable configuration files, e.g., files created for a user-selectable typical hardware operation, e.g., analysis of a portion or an entire genome, e.g., ancestry analysis, diagnostics of diseases, search for new drugs, profiling of proteins, etc. Preset parameters of this type, for example for performing similar analyses, can be stored in a library. For example, if the platform described herein is used for, for example, cancer research, the predefined parameters may be configured differently than if the platform is aimed at genealogical research.

[00454] More specifically, in oncology, accuracy may be an important factor, and therefore system parameters may be set to provide increased accuracy at the possible cost of speed. However, speed may be a key determining factor for other genomics applications and, therefore, system parameters may be set to maximize speed, which may, however, come at the expense of accuracy. Accordingly, in various embodiments, parameter settings frequently used to perform various tasks may be preset in the library to provide ease of use. These parameter settings may also contain information about software applications that need to be enabled when the system starts. For example, such a library could contain the code that runs the API, and subsequently example files, batch files, and any supporting information needed to run the system. In this way, the library can be configured to compile software to run both the API and various executable programs.

[00455] In various cases, PCI 104 and/or chip 100 may also include a memory unit, such as random access memory (RAM) or dynamic random access memory (DRAM) with, for example, a DDR3 interface, and such memory may be used to provide the performance of the various modules described in this document, for example, the mapping module, the leveler, and/or the sorter. For example, DRAM may be used where a check sequence, a lookup table, and/or a lookup table pointer, and/or readable chunks can be stored. Next, as shown in FIG. 9., the memory may be used to support the operation of various other modules, for example 114, described herein, such as, for example, a duplicate elimination module, a local realignment module, a base quality recalibration module, a variant search module, compression modules and/or unpacking. For example, DRAM may be used where sorted reads, annotated reads, compressed reads, and/or genetic variants can be stored. Further, the memory configuration may be configured to include a separate interface for each of the various memory modules used by the aligner and/or any other module, such that each memory module may comprise a file level and a logical level. As discussed above, since multiple memory units and/or multiple modules may be involved, a chip-level interconnect may be included to allow information to be exchanged across the chip 100.

[00456] Thus, in various cases, the apparatus described in the present invention may include a chip 100, where such chip includes an integrated circuit formed by a group of permanently connected digital logic circuits that can be interconnected by one or more electrical interconnections. In various embodiments, one or more physical electrical interconnects includes an input element by which the integrated circuit can couple to an electronic data source to receive data. Further, in various embodiments, the permanently connected digital logic circuits may be organized into a set of processing devices, for example, where each processing device may be constituted by a subset of permanently connected digital logic circuits that are configured to perform one or more digital steps. sequence analysis pipeline. More specifically, each subset of permanently connected digital logic circuits is configured to perform one or more stages of a digital sequence analysis pipeline.

[00457] In various cases, a group of processing devices may include one or more mapping modules 112, alignment modules 113, and/or sorting modules 114a, for example, such that one or more of these modules are in a mounting configuration. For example, a mapping module may be included in a mounting configuration, and such a mapping module may access an index of, for example, one or more reference genetic sequences, for example, from memory, for example, through one or more sets of physical electronic interconnects, for example , for mapping a set of reads from one or more segments of one or more reference genetic sequences. Further, in various cases, an alignment module may be included, and such an alignment module may access one or more reference genetic sequences, for example, from memory, for example, through one or more sets of physical electronic interconnects, for example, to align a set of reads fragments of one or more segments of one or more control genetic sequences. Moreover, in various cases, a sorting module may be included, and such a sorting module may access one or more aligned sequences, for example, from memory, for example, through one or more sets of physical electronic interconnects, for example, to sort a set of reads fragments of one or more chromosomes of one or more control genetic sequences. Likewise, in various cases, the chip may comprise one or more of local realignment, duplicate marking, base quality recalibration, and/or variant detection modules, for example, in a mounting configuration in the same manner as the modules described above, to perform the corresponding functions.

[00458] As stated above, in various cases, the configuration of one or more integrated circuits according to the present invention may be configured for one or more chips, such as an FGPA, an ASIC, and/or a fabric chip ASIC. For example, an integrated circuit, by type, is a collection of electronic circuits arranged on a small wafer or “chip” of semiconductor material, such as silicon. Integrated circuits typically include circuit elements that can be integrally coupled with them and interconnected electrically. A prototypical digital integrated circuit includes a variety of elements, such as, for example, one or more logic gates, flip-flops, hubs, and other various elements, configured or customized to operate the circuit as a microprocessor or other microcontroller, such as for binary signal processing "zeros" and "ones", for example, when performing one or more operations according to the present invention.

[00459] More specifically, one or more mask-programmable logic gates may be configured or programmed to perform logical operations, such as Boolean functions of one or more input logic variables, so as to produce a single output logic variable. Such logic gates can be configured using one or more diodes or transistors so that the gates act as electronic switches. In various cases, logic gates can be cascaded, similar to how Boolean functions can be arranged, which thus makes it possible to build a physical model of any Boolean logic and, therefore, any algorithms and any mathematical models that can be described by Boolean logic, such as those described herein, and can be implemented by the logic gates of the integrated circuits of the present invention. In various embodiments, a set of gates may be located on a semiconductor wafer, thereby forming a gate array, such as a gate array circuit.

[00460] In various cases, the integrated circuit may also include one or more flip-flops. The trigger may be a circuit, or at least a portion of the circuit, configured as a valve. Typically a flip-flop has two stable states and can change from one to the other, for example, by a signal applied to one or more control inputs, and therefore the flip-flop will have one or two outputs. Triggers are used to store state information and therefore can be used as basic storage elements, such as in sequential logic operations. The integrated circuit may also include a hub. The hub may be configured to select one of several input signals, such as digital (or analog) input signals, and may further be configured to relay the selected input signal to an output. Thus, a hub can be used to increase the amount of data sent over a network over a given period of time for a given bandwidth.

[00461] In some cases mentioned herein, a typical integrated circuit may include any number, from one to millions, of such elements configured to perform operations such as those described herein, where the various elements occupy only a few square millimeters. The small size of such circuits provides high speed, low power dissipation and reduced manufacturing costs.

[00462] Such integrated circuits can be manufactured using a wide range of different technologies, however, they are typically manufactured as a monolithic integrated circuit. For example, a typical integrated circuit, such as a semiconductor, may be manufactured using a layer-by-layer process, such as a process that includes three major steps such as image transfer, placement, and etching. In various cases, one or more of these steps may be supplemented by further processing steps such as alloying, purification and the like. For example, a typical manufacturing procedure may involve a semiconductor wafer, such as a monocrystalline silicon wafer, used as a substrate on which the integrated circuit is assembled or, for example, printed. Photolithography is then used to print onto the semiconductor wafer, for example to mark out different areas of the substrate, which can then be doped and/or on which conductive traces, for example metal, such as aluminum, will be printed.

[00463] Typically, an integrated circuit consists of one or more sets of overlapping layers such that the structure of each layer is determined by photolithography. Some layers may form diffusion layers that mark the locations where various dopants have diffused into the substrate, and other layers may define locations where additional ions can be implanted. Additional layers may be configured to be conductive (eg, polysilicon, metal, or the like) or to provide connections between conductive layers. For example, a transistor may be formed at one of the locations where a gate layer (polysilicon or metal) intersects the diffusion layer, and, in various cases, curved strips may be used to form on-chip resistors. Typical integrated circuits may include: ASIC, FGPA, and/or structured ASIC.

[00464] Integrated circuits are often manufactured as general purpose circuits. However, in various cases, such as some of those described herein, the integrated circuit may be application-specific, such as an application-specific integrated circuit, or “ASIC.” Typically, an ASIC belongs to the class of “standard ASICs,” that is, integrated circuits specialized for a specific, not universal, application. Typically an ASIC may contain a large number of logic gates, such as in some cases over 100 million gates, which may be configured to perform a large number of different operations, such as configured to act as microprocessors and/or memory units including ROM, RAM, EEPROM, flash memory, and other large building blocks, such as those designed to perform the operations described in this document. What's unique about an ASIC is that because it is a chip designed to perform a specific set of application tasks, it can be manufactured in a way that makes it specialized, such as using a gate array protocol design.

[00465] For example, a gate array or unswitched logic array (ULA) can be used in the design and manufacture of application-specific integrated circuits (ASICs). In such a case, the ASIC may be manufactured from a prefabricated chip containing active devices such as gates, such as NAND gates, which may not be interconnected initially, but are subsequently interconnected, for example, according to a gate array design protocol. for example, using an additional metal layer, for example in a manufacturing plant. Thus, with regard to ASIC manufacturing, a gate array may be pre-fabricated on a silicon chip board, which as a result has no specific function, but contains one or more transistors, standard NAND or NOR logic gates, and may further contain other active devices located at specified locations on the semiconductor wafer, which in this case may be called the "base chip". Thus, the creation of a circuit that performs certain specialized functions may be completed by adding a final surface layer or layers of metal interconnects to the chips on the base chip in the final stages of the manufacturing process, and connecting elements that provide the required specialty of the chip, for example, according to a design protocol.

[00466] More specifically, the gate array design protocol involves the use of a fabrication method in which various diffusion layers, for example with transistors and other active elements such as those described above, are ready for use and built on general purpose semiconductor wafers, and before performing the step metallizations are in such a state that the various elements remain unconnected. In such a case, the chip may be further specialized at some stage in accordance with various parameters associated with its specific purpose, for example, at the physical design stage, when the interconnections of the finished device are determined. For example, base gate array dies are usually prefabricated and stored in a warehouse in large quantities awaiting the specialization stage. A circuit intended for a specific application must be assembled on a gate array in such a way that the circuit has a sufficient number of gates, wiring connections, and input/output (I/O) pins to perform the required functions.

[00467] Since requirements may vary, gate array wafers are often produced in standard series, both larger products, which have, for example, more resources, but are correspondingly more expensive, and relatively smaller products, which have a limited choice of resources, but are less expensive. The selection of a suitable wafer standard should be based on the amount of resources required to perform the selected functions. The amount of resources that will need to be deployed can be determined relatively easily, for example by counting the number of I/O pins required, however, the number of wiring paths can vary within some limits, and therefore must be chosen with care. However, since the base die is typically prefabricated, design and manufacturing, as required by a specific design protocol, can be completed in less time than a standard partial or full customization (FPGA) design. Also, this gate array approach reduces template manufacturing costs, since fewer templates need to be manufactured. In addition, the time and cost of performing instrumentation testing is also reduced because the same contact fixtures can be used for all gate arrays manufactured with the same die size.

[00468] In such a case, one manufactured standard chip, for example, an ASIC, may consist of any number from two to nine, or ten, or twelve or more layers, such that one or more, for example, all going one behind another, metal layers are arranged perpendicular to those underneath. Such fabrication methods are convenient because they produce a relatively specialized chip design in a relatively short assembly time, since the final metallization process can be performed quickly. However, such gate array chips, such as ASICs, are often suboptimal because when the design is placed on an “off-the-shelf” semiconductor wafer, the provided resources are typically not 100% utilized. Another disadvantage of ASICs is non-recurring engineering costs (NRE), which can run up to millions of dollars. However, unit costs can be relatively low for ASICs.

[00469] An alternative to a standard ASIC for custom chip manufacturing is a field programmable gate array, or “FPGA.” FPGAs use rewritable programmable logic blocks and interconnects that allow the FPGA to be designed or at least partially redesigned for use in different application problems or in the same application problem, but solved in many different ways at different points in time. More specifically, a field programmable gate array is an integrated circuit that is designed to be configured one or more times, for example by the customer or designer, for example, after manufacture.

[00470] Typically, FPGAs have a large number of resources in the form of logic gates and/or memory blocks, such as RAM, blocks that can be configured to implement complex digital calculations. For example, FPGAs contain both programmable logic components, called “logic blocks,” and a large number, such as a hierarchical structure, of reconfigurable interconnects that allow these blocks to be in a “wired” state. More specifically, FGPAs can contain a large number of variable logic gates that can be arranged in a variety of configurations, thereby forming logic blocks that can be configured to perform a wide range of complex combinational functions, such as those related to operations, detailed in this document. In various cases, FPGA logic blocks can be configured to include memory elements, such as something as simple as flip-flops, or more complete memory blocks like ROM or RAM. Because FPGAs use very fast I/O and bidirectional data buses, validating valid data timing samples and hold times can be difficult in some cases. Thus, in some cases, proper layering may allow resources to be allocated to the FPGA to accommodate timing constraints. An FPGA can therefore be used to implement any logic function that an ASIC can perform. However, the ability to update and adjust functionality after delivery, partially reconfigure part of the structure, and low one-time design costs in relation to ASIC (despite the higher cost per unit of production), determine its advantages for many applications.

[00471] In some cases, the fabrication of a typical FPGA uses a coarse-grained architecture approach to combine the logic blocks and interconnects of a traditional FPGA with embedded microprocessors and associated peripheral devices to create a complete "system on a programmable chip". In some cases, the FPGAs described in the present invention may be reprogrammable "at run time" which, in accordance with the methods set forth herein, may enable reconfigurable computations or the creation of reconfigurable systems, such as CPUs, which can reconfigure itself to become compatible with the operations described in this document. In some cases, microprocessors with configurable software may be used to form a matrix of processor cores and FPGA-like programmable cores that may reside on a single chip.

[00472] A typical FPGA architecture may include an array of configurable logic blocks, I/O pads, and/or one or more trace channels. Typically, a logic block may include one or many logic cells, and a typical cell may contain a four-input lookup table (LUT), a three-input full adder (FA), and/or a flip-flop, and something else. similar, functioning to generate an output signal. In various cases, the output can be either synchronous or asynchronous. The application circuit may be implemented in an FPGA, and the number of logic blocks, I/O, and wiring traces that must be included in the circuit, which may vary, can be determined according to the design documentation. It should be noted that since unused traces can add cost and reduce the performance of an integrated circuit without providing any benefit, the number of conductive traces must be sufficient, that is, selected to support the layout of lookup tables (LUTs) and I/O, but not in excess. Further, since synchronization signals typically travel over a dedicated routed network (eg, a global buffer), such signals can be managed separately.

[00473] The FPGA described herein may also contain elements that provide higher-level functionality implemented in the circuit, such as one or more multipliers, generic digital signal processing (DSP) card assemblies, embedded processors, high-speed gates I /O, and/or built-in memory blocks. Including these generic functions embedded on a wafer reduces the required area and allows for increased speed of execution of such functions. It should be noted that the described FPGAs can be used to verify the performance of systems, including at the pre- and post-manufacturing stages and the firmware development stage, for example, to verify the functionality of the final design before production of chips such as a standard ASIC or structural ASIC, which may represent the final product.

[00474] When manufacturing a generic integrated circuit, such as an FPGA or the like, having the desired functionality described herein, one or more of the following steps may be performed in any logical sequence. The first step may be hardware description language (HDL) or circuit design. Next, a computer-aided design system, such as CAD, can be used to create the process netlist. The netlist can then be written directly into the FPGA architecture, for example through a "place and route" process, using appropriate software to place components and route connections. Once the design and functionality are completed, the generated binary file can be used to configure/reconfigure the FPGA.

[00475] In a typical design protocol, design performance simulations may be performed at various design stages. First, the description of the register transfer logic, RTL, such as VHDL or Verilog, can be simulated by creating test benches to simulate the system and observing the results. In some cases, the synthesizer may convert the intended design into a netlist, and after the synthesizer has finished converting the intended design into a netlist, the netlist may be translated into gate-level description form. At this stage, the simulation can be performed, for example, again, in order to ensure that the synthesis was carried out without errors. The design then enters the FPGA placement stage, where signal propagation delays can be added and the simulation can be run, for example, again, with this data added to the netlist, for example, before final performance testing and further manufacturing, e.g. one or more ASICs or structured ASIC chips.

[00476] Thus, a structured ASIC is a combination of an ASIC and an FPGA, that is, it is between an FPGA and an ASIC. The traditional "standard ASIC" described above is usually expensive and requires significant development time. For example, when developing a standard ASIC, it may be necessary to produce a large number of photolithographic templates for each standard ASIC design separately. However, after this initial development investment, typical production costs become very low and operating parameters related to power, frequency and performance can be easily optimized.

[00477] Unlike standard ASICs, typical FPGAs and/or CLPDs containing programmable logic blocks are relatively inexpensive and fast to develop, primarily because existing devices are electronically programmable and photolithographic patterns are not required. However, in terms of performance parameters related to power, frequency and performance, such devices are significantly inferior to standard ASICs, and their unit costs can be very high, especially for high-performance devices.

[00478] A structured ASIC, on the other hand, is a compromise between the two devices mentioned above. Unlike gate arrays, structured ASICs typically contain ready-made configurable memory and/or analog blocks. Thus, their development cost is significantly lower compared to standard ASICs, since only a few photolithographic templates need to be produced for each individual structural ASIC design, such as configurable metal layers. And, despite the fact that the cost per unit is significantly higher than for standard ASICs, it is nevertheless significantly lower than for FPGAs. In terms of power and frequency, the structured ASIC occupies an intermediate position between standard ASICs and FPGAs, however, in terms of performance it is close to the largest FPGAs. Thus, in many cases, a structured ASIC can be a technology solution that reduces the initial cost and time required to develop a new application specific integrated circuit.

[00479] With respect to structural ASIC design and fabrication, before volume development of structural ASICs begins, a "base die" may be designed, for example, using standard ASIC methodology. As stated above, the base die may include typical integrated circuit layers, such as one or more transistors, memory blocks or cells, input/output cells, phase-locked loops or other clock oscillators, or the like. If necessary, the base chip may contain flip-flops, latching relay elements, and/or multi-transistor combinational gates. The base die may contain a number of wiring connections between components, but not necessarily a complete set of them to implement a complete logic circuit, since connections can be added later. Note that the base die could theoretically be designed to include standard ASICs, potentially containing large complex modules, and the operating parameters (power, frequency, performance) of the base die could be optimized to be the same as for standard ASICs. Photolithographic templates can be produced for the contents of the base die, with the number of templates being approximately the same or less than with a standard ASIC. Thus, the base die includes a set of digital logic circuits that may or may not be permanently connected for a specific device function.

[00480] Once the base die is designed, a series of one or more structured ASICs can be manufactured, for example, built on the same base die. Typically, designing multiple fabric ASICs uses the resources of a single base die to amortize the cost of the base die across multiple designs. Each individual structural ASIC design can be implemented by a set of new wiring connections between components (transistors, etc.) in the base chip, which will provide the construction of higher-level gates, flip-flops, latching relay elements, memory blocks, and large logic modules. Accordingly, the wiring connections thus defined can be implemented using a smaller number of additional "configurable" metal layers 904A and 904B built on the base die, for example, by connecting pads or across the base die, for example, using wiring connections on the configurable metal layers. These additional metal layers are called “configurable” layers because they can be specialized for any ASIC design; however, they are fixed during manufacturing and cannot be physically rewired, but only using the logic circuit implemented in the device. The number of configurable metal layers can be any.

[00481] Most of the envisioned logic circuits can thus be implemented using a base die and proper wiring connections of the metal layers, since the base die contains sufficient logic resources (transistors, memories, etc.) to form all the necessary elements logic circuit. The number of configurable metal layers may vary from one structural ASIC design to another, but can typically range from approximately 1 to 5 configurable layers. A small number of photolithographic templates can be produced depending on the number of configurable metal layers, and during device fabrication, the complete set of templates (base die templates and configurable metal layer templates) can be used to fabricate the semiconductor wafers of the finished ASIC die. In other cases, the base die wafers could be mass-fabricated in advance and the metal layers added to the finished wafers of specialized structural ASIC designs at a later stage of fabrication.

[00482] Preferably, the structural ASIC base die is designed in one step, for example by a first designer, while the logic circuits of the structural ASICs built on that base die can be designed in a second step, for example, by different designers using the work of the structure ASIC designer. In particular, various workers may typically be responsible for the "client side" of the logic circuit, specialized for the required functionality of the integrated circuit, for example, RTL (register transfer logic), code development, performance modeling, emulation, regression testing, debugging, and the like; whereas the structural ASIC designer may be responsible for the design of the “behind the scenes”, including synthesis, placement and routing of interconnects, analysis of static failure risks, embedding test logic, and/or handover to production. Additionally, microelectronic manufacturing may be involved, for example, to produce photolithographic templates and wafers, and/or test and/or package devices. In various cases, the structural ASIC designer may also design specialized base dies for a particular class of application, such as those containing logic resources of the type and quantity specialized for that application.

[00483] Thus, by having prefabricated metal layers (which reduces manufacturing time) and predefined characteristics of what is on the wafer, such as the base die (which reduces design time), design and full manufacturing cycles can be completed in less time time in case of structured ASIC compared to conventional ASIC production. For example, when designing a standard ASIC or FPGA, such as a gate array, the user may often need to design the power, clock, and test structures themselves. However, in the case of structured ASICs, they can be off-the-shelf, which can reduce production time and costs compared to standard ASICs and gate arrays.

[00484] In particular, the design challenge for structural ASICs is to place the circuitry in a fixed arrangement of given cells. More specifically, a comparative structure ASIC architecture may generally include two main layers, such as a structure element level and a structure element matrix level. Such building blocks may include both combinational and sequential functional blocks, which can function as either logic or storage elements. Additionally, with respect to structure element matrices, uniform or heterogeneous matrix styles may be used, for example, in a fixed arrangement of structure elements.

[00485] As a result, when designing a structural ASIC, the logical template layers can be defined in advance. In such a case, design modifications and specialization can be accomplished, for example, by creating specialized metal layers that create specialized connections between predefined lower-level logic elements. Also, structural ASIC design tools can be significantly lower in cost and easier (faster) to use than standard cell design tools, since it is not necessary to perform all the functions that are performed using standard cell design tools. More specifically, existing CAD for standard cells can be used in the design. In some cases, however, CADs specifically designed for structural ASICs may be used. Specialized installation tools may also be used. Further, as described in this document, new and improved algorithms have been developed using the modular design of structural ASICs with improved timing control. Additionally, the methods described herein can be used to improve quantification and analysis processes, as noted above.

[00486] Thus, a structured ASIC may be a technology intermediate between field programmable gate array designs and standard ASICs. More specifically, because only a small number of specialized chip layers are required, one-time design costs (NRE) can be significantly lower than for "standard cell" chips or "fully customized" chips, where a full set of patterns must be fabricated for each design. Thus, a structured ASIC provides high performance (same as a typical ASIC) at a low NRE cost (same as an FPGA). Thus, the structured ASIC manufacturing process provides a fast-to-market product that is relatively easy to develop and has a reduced cost.

[00487] In some cases, however, it is preferable to use an FPGA in which the interconnects and logic blocks are programmable after manufacturing. This provides greater design adaptability and ease of debugging during prototyping. However, the ability to build large circuits in size and speed on an FPGA is often limited, in some cases due to the inherent complexity of programmable interconnect routing and/or the significant amount of space occupied by the various software elements included within it. On the other hand, ASICs have a number of disadvantages, such as an expensive development process, due in part to the fact that each new design usually requires a complete set of new templates. A structured ASIC may therefore be the optimal solution when choosing between the two designs mentioned above. It may have the same basic structure as an FPGA, but be pattern programmable like an ASIC instead of user programmable by adjusting the configuration of the layers in the through holes between the metal layers. For example, one or more, for example, each static random access memory (SRAM) configuration bit may be replaced by whether or not the configuration includes an end-to-end or inter-group metal contact.

[00488] For example, with respect to structured ASIC architecture, a typical architecture can often be fine-grained, medium-grained, and/or hierarchical. A fine-grained architecture may include a large number of connections at the input and output of a structural element, while larger structures have fewer connections to the structural element, and may degrade the functionality it supports. Each individual design will have its own set of advantages depending on its structural features. More specifically, a fine-grained architecture may include structural elements containing disparate, unconnected components, such as transistors, resistors, and other control elements, that can be subsequently connected. The middle-structure element architecture can include both generic logic and gates, MUXs, LUTs, and/or storage elements such as flip-flops. In contrast, a hierarchical architecture may include mini-structural elements that, for example, include gates, MUXes, and LUTs, but typically do not include typical storage elements such as flip-flops. In other cases, a mini-element may be combined with registers or flip-flops.

[00489] In terms of implementing a structural ASIC, various manufacturing steps may include one or more register transfer logic (RTL) design steps; logic synthesis for transferring RTL to structural elements; embedding test functions for improved testability and fault detection; installation for transferring each structural element to the matrix element and placing each of the elements in a fixed order; physical synthesis for improved timing of time intervals on the circuit and optimized placement of each element; Clock synchronization synthesis to distribute the clock synchronization network and minimize clock distortion and delays; laying interconnections or, in other words, installing wiring connections between various elements. In different cases, such steps may be performed in a different logical order, so as to make the design process, such as that related to logic synthesis, less complex, and also to assemble a more complete ready-made library of structured ASICs that contains an expanded view of what exactly can be implemented in accordance with the project.

[00490] Moreover, a common practice in the development of processor cores is to license the design of the processor as it is built by various users into their own semiconductor devices. The list of such embedded cores may include general purpose processors such as ARM, PowerPC, Krait, etc., as well as more specialized processors such as graphics processing units (GPUs) or vector processors. Embedded processor cores can be large, complex pipelined logic units operating at high operating frequencies, such as 1 or 2 GHz to about 3 or 6 GHz or higher. Achieving such high frequencies of processor cores and associated memory buffers requires precise physical placement and interconnect routing, and as a result, embedded processor technology can often be accompanied by a "hardware macro" (for example, to determine the precise placement and interconnect routing of subcomponents) , designed for a specific manufacturing process.

[00491] However, such an embedded processor core may not be optimal when implementing a structural ASIC using configurable metal layers. Hardware macros are not typically used for configurable fabric ASICs, and even if an embedded processor were implemented as closely as possible to match its hardware macros in configurable metal layers, it would likely be frequency limited (e.g. 30% or 50% of the nominal operating frequency), and most likely consumed very large amounts of the available resources of the base crystal. The relatively small area of a fabricated ASIC compared to a standard cell could result in an embedded processor occupying significantly more physically available area on the silicon die, and, in combination with the reduced operating frequency, the performance per unit area (or cost) could be lower than when constructing a standard cell with the same embedded core.

[00492] However, embedding one or more processor cores into a fabric ASIC base die is feasible, for example, using the standard cell design methodology described herein, which includes the use of a hardware macro. This would allow maintaining the operating frequency, full performance and normal footprint. The processor core and/or buffer memory input and output wiring could be interconnected to other base die resources or, preferably, embedded in a configurable metal layer layout, allowing the embedded cores to be interconnected with any infrastructure and logic modules implemented on each specific chip. structural ASIC designs. Thus, the embedded processor cores become a resource of the base crystal, available for the further implementation of many different structural ASIC projects using this base crystal.

[00493] Embedded within a fabric ASIC, processor cores may be coupled to a logic infrastructure such that software (firmware) running on the cores can access, use, and communicate with various memory blocks and other on- and off-chip resources. information with one or all other logic modules on the chip through memory blocks and/or directly. Thus, processor cores can work in parallel with other logical modules and/or interact with other logical modules to perform cooperative work, for example, when processor cores send a request to other modules to perform tasks, or other modules send a request to processor cores to perform tasks, or both both.

[00494] When bioinformatics (Bio-IT) accelerating modules (e.g., mapping, alignment, sorting, duplicate labeling, base quality recalibration, local realignment, compression, decompression, etc., described herein) are implemented on the FPGA and /or structured ASIC with embedded processor cores, the resulting system on a chip (SOC) has a number of important advantages, especially when combined with speed and adaptability. The greatest speed can be achieved using hardware accelerator modules, and the greatest adaptability can be achieved through the full programmability of processor cores. By reprogramming the processor cores, Bio-IT's executable algorithms can be easily modified, and these algorithms can run orders of magnitude faster than traditional CPUs because the computation-intensive operations can be offloaded to hardware accelerators. Information exchange and memory organization can be optimized for the processor and accelerator to work together. Additional acceleration of software algorithms can be achieved with the help of additional hardware modules designed for pre- and post-processing of data used by processor cores, for example, to convert data overlapping readable fragments of a control genome region into an overlay data structure for representation in processor cores. On some architectures, the instruction set can be extended to take advantage of hardware resources; New processor instructions can be defined in the Bio-IT SOC to access Bio-IT hardware acceleration features.

[00495] As presented in Table II below, a structured ASIC therefore has several prefabricated advantages over, for example, an ASIC or an FPGA. For example, various components can be "almost" connected, for example, in a variety of off-the-shelf configurations, and multiple central and local clock generators can be prefabricated. This means, therefore, that problems with signal integrity and timing must be essentially resolved. Additionally, only a few metal layers need to be produced. Further, unlike standard FPGAs, a structured ASIC must have power, performance, and power consumption parameters close to a standard ASIC. This allows for a faster and easier development process, as well as lower non-recurring engineering (NRE) costs compared to standard ASICs, and significantly reduces the time from device development to release. Moreover, there should be no need to solve the problem of distortion.

TABLE II FPGA Structural ASIC Standard ASIC Crystal area Very big Small Very little Power resource usage High Low Very low Operating frequency Low High High Logical capacity Average Average High Development cost Very low Low High Unit cost Very high Low Very low

[00496] The structured ASIC therefore has several advantageous properties, including one or more of the following: low one-time design costs, simplified development requirements, reduced template manufacturing costs compared to a standard ASIC , with the added benefits of high performance, low power consumption, fewer layers to fabricate, and less complexity in circuit-ready configuration of prefabricated cell blocks, resulting in reduced fabrication time. However, the structured ASIC has some disadvantages, such as the lack of adequate design tools, which can be expensive and are achieved by modifying traditional ASIC tools. Further, such new architectures still require formal evaluation and comparative analysis. And there are problems of choosing the optimal lookup table between 3, 4, and 5 input LUTs and/or the optimal size of distributed RAM.

[00497] Thus, taking into account the above, ASIC, FPGA and structured ASIC have both their advantages and disadvantages. For example, a standard ASIC is difficult to design, requires more development time, and has high NRE costs. However, the ASIC can support large designs, complex designs, and has high performance with low power consumption, which can therefore result in low cost per unit (at high volumes). On the other hand, FPGA is easy to design, requires less development time, and has low NRE cost. However, FPGAs may have limited design size and/or complexity, limited performance, and high power consumption, which may result in high unit costs. In many cases, a structured ASIC can be designed to maximize specified benefits and reduce specified disadvantages. For example, in general, the ratio for the number of gates per unit area can be approximately expressed as 100:33:1 for standard ASIC, fabric ASIC and FPGA; for performance (based on clock speed) - as 100:75:15; for power - like 1:3:12.

[00498] As discussed above, in various cases the chip 100 of the present invention may be configured as an expansion card, for example, if the chip includes a PCIe bus and is located to communicate with one or more memory blocks, for example, surrounded by memory blocks , for example, is largely surrounded by memory blocks, for example, completely surrounded by memory blocks. In various embodiments, the chip may be a high-density FPGA and/or a high-speed FPGA, which in various cases may be converted to an ASIC or structured ASIC. In various cases, the chip may be a structured ASIC, which may be converted to an ASIC. In some cases, the chip may be an ASIC.

[00499] As discussed above, the modules described herein may be implemented in chip hardware, for example, permanently coupled thereto, and in such cases, the implementation may result in their operation occurring at a higher speed compared to a software implementation in such a way that, for example, a minimum number of instructions are called, read and/or executed. Thus, with such a unique implementation in hardware, the modules described in the present invention can function directly in accordance with their operating parameters, for example, without the need to invoke, read and/or execute instructions. Additionally, memory and processing time requirements can be simplified, for example, when on-chip information exchange occurs through files rather than through memory access. Of course, in some cases, the chip and/or board can be increased in size, for example to accommodate more on-chip memory, so as to increase parallel processing capabilities, which can result in even greater processing speeds. For example, in some embodiments, the chip described in the present invention may include on-chip DRAM so that the chip is not dependent on external memory, which would therefore further increase processing speed, for example, when running the Burrows algorithm. Wheeler instead of the arrangement table and the shuffling function, which may, in various cases, depend on external, for example, central memory. In such cases, the entire software pipeline can be completed in 6 minutes or less, for example, from its start to its end.

[00500] As discussed above and illustrated in FIG. 8, there are various points at which a given module may be located in hardware, or located remotely, such as on a cloud-accessible server. If this module is located on a chip, for example permanently connected to the chip, its function may be performed in hardware, however, if required, the module may be located remotely, in which case the platform may include the necessary set of tools for sending the relevant data to a server, for example a cloud-accessible server, so that the specific functionality of the module can be used to further process the data in accordance with the protocol selected by the user. Thus, a portion of the platform may include an Internet remote access interface for performing one or more tasks related to the operation of one or more modules described herein. For example, if mapping 112, alignment 113, and/or sorting 114a modules are all on-chip modules, then, in various cases, one or more of local realignment 114d, duplicate marking 114b, base quality recalibration 114c, and/or search options 115 may be on the cloud.

[00501] In particular, after the genetic data has been generated and/or processed, for example, using one of the primary and/or secondary protocols, for example, mapped, aligned, and/or sorted, for example, to create one or more files determining variants, for example, to determine how the genetic sequence data of a subject of interest differs from one or more reference sequences, a further aspect of the invention may be associated with performing one or more analytical functions on the generated and/or processed genetic data, such as further, e.g., tertiary,processing. For example, the system configuration shown in FIG. 8-11 may be configured to further process the generated data and/or data resulting from secondary processing, for example, to run through one or more tertiary processing software pipelines 700, for example, through one or more genome software pipelines, software pipelines epigenome, metagenome software pipelines, co-genotyping, MuTect2 software pipelines, or other tertiary processing software pipelines, such as the devices and methods described herein. For example, in various cases, an additional level of processing 122 may be provided, for example, for disease diagnosis, therapeutic treatment and/or prophylactic actions, for example, including NIPT, NICU, Cancer, LDT, AgBio and other similar methods for disease diagnosis, prevention, and /or treatments that involve data resulting from the operation of one or more such primary and/or secondary and/or tertiary software pipelines. Thus, the devices and methods described herein can be used to generate genetic sequence data, which can then be used to generate one or more variant definition files and/or other related data, which can then be the subject of research by others. run tertiary processing software pipelines in accordance with the devices and methods described herein, for example, both for the diagnosis of a specific and/or systemic disease and for prevention and/or therapeutic treatments and/or techniques under development.

[00502] As described above, system 1 presented herein may include generating, for example, using an on-chip sequencer technology solution described herein, or in other cases, obtaining genetic sequence data, and may include performing one or more secondary processing protocols, for example, one or more of the following: mapping, alignment, sorting of the generated genetic sequence data, for example, to obtain one or more variant lookup files, for example, to determine how The genetic sequence data of one study subject differs from one or more reference genome sequences. A further feature of the invention may be related to the performance of one or more analytical functions in relation to the obtained and/or processed genetic data, such as further, for example tertiary, processing, which may be performed on the same chip or in conjunction with the same chip or chipset, on which the above-mentioned sequencer is located.

[00503] In the first case, for example, regarding the generation, acquisition and/or transmission of genetic sequence data, as shown in FIG. 8, such data may be received locally or remotely and/or the result may then be processed, for example, using local computing resources 100, or may be transferred to a remote location, such as remote computing resources 300, for further processing. For example, the generated genetic sequence data can be processed locally and directly, for example, if one chipset and/or one device has sequencing and secondary processing functionality. The generated genetic sequence data can also be processed locally, but not directly, if the sequencing and secondary processing functionality is located on different machines that are located in the same room, but separated in space and yet connected with signal transmission capability, for example through a local network 100. In other cases, genetic sequence data may be processed remotely, for example, using NGS, and the results may be transmitted via cloud network 50 to a remote location, for example, geographically distinct from the sequencer.

[00504] Specifically, as shown in FIG. 8-11, in various embodiments, a nucleotide sequencer may be provided directly in the workspace, such as an on-chip sequencer or NGS, where said sequencer is coupled to a local computing resource directly or indirectly, such as a connection through a local area network 10. Local computing resource 100 may include or, in other cases, be combined with one or more mechanisms for generating data 110 and/or receiving data 120. Such mechanisms can be any mechanisms configured to generate and/or, in other cases, receive data, e.g. , analog, digital or electromagnetic data related to one or more genetic sequences of the object or group of objects under study.

[00505] For example, such data generation engine 110 may be a front-end processing device, such as a sequencer such as NGS, an on-chip sequencer, or other similar mechanism for generating genetic sequence information. Further, such data acquisition engine 120 may be any engine configured to receive data, such as generated genetic sequence information, and/or, when used in conjunction with data generation engine 110 and/or computing resources 150, capable of utilizing one or more secondary processing protocols, such as a secondary processing software pipeline configured to execute mapping, alignment, sorting, and/or variant search protocols on generated or received sequence data as described herein. In various cases, the data generating apparatus 110 and/or data receiving 120 may be connected, for example, through a local network 10, for example, for local storage 200, or may be connected through a cloud network 30, for example, for transmitting and/or receiving data such as as digital data related to the primary and/or secondary processing of genetic sequence information, for example, to or from a remote location 30, such as for remote processing 300 and/or storage 400. In various embodiments, one or more of such components may be interfaced with signal transmission capability using a hybrid network as described herein.

[00506] Local computing resource 100 may also include, or in other cases be combined with, a compiler 130 and/or a processor 150 such that the compiler 130 is configured to compile generated and/or received data and/or associated data and the processor 150 is configured to process the generated and/or received and/or compiled data and/or control the system 1 and its components, as described herein. Further, local computing resources 100 may include a data compression device 160 configured to compress data, such as generated and/or received data and/or data resulting from primary and/or secondary processing, which may be compressed, for example, before sending via local network 10 and/or cloud network 30 and/or hybrid cloud network 50.

[00507] In specific examples, as can be seen in FIGS. 8-11, system 1 may be configured to perform further processing of generated and/or secondarily processed data, such as through local 100 and/or remote computing resources, such as by running it through one or more tertiary processing pipelines, such as one or more of a genome pipeline, an epigenome pipeline, a metagenome pipeline, a co-genotyping pipeline, a MuTect2 pipeline, or another tertiary processing pipeline. Such data may then be compressed and/or locally 200 locally and/or transferred for remote storage.

[00508] In further examples, system 1 may comprise an additional layer of processing modules, such as those configured to perform additional processing, such as diagnosis, disease and/or therapeutic property detection, and/or prevention thereof. For example, in various examples, additional levels of processing may be provided, for example, for disease diagnosis, therapeutic treatment and/or prophylactic prevention, including, for example, non-invasive prenatal testing (NIPT), neonatal intensive care unit (NICU), cancer, performed in the laboratory research (LDT), agrobiology (AgBio) and other types of diagnosis, prevention and/or treatment of such diseases that use data generated by one or more of the specified primary, secondary and/or tertiary pipelines.

[00509] Accordingly, presented herein is a system 1 for obtaining and using a global hybrid cloud network 50. For example, according to the present document, the cloud 30 is primarily used for storage, such as at a remote storage location 400. In such an example, data computation is performed locally 100 through local computing resources 150, and in the case where storage requirements are significant, cloud 30 is accessed to store data generated by local computing resources 150, for example using remote storage resources 400. Consequently, the generated data is typically either managed entirely locally 100 or entirely remotely 300 in the cloud 30.

[00510] In particular, in a general embodiment of a bioinformatics analysis platform, the computation 150 and/or storage 200 functions are performed locally, and when storage requirements exceed the overall storage capacity, or when there is a need to make stored data accessible to other remote users, such the data may be transmitted via the Internet 30 to the cloud for remote storage 400 therein. In such an example, when the computing resources 150 required to perform the computing functions are minimal and the storage requirements are significant, the computing function 150 may be executed locally 100 and the storage function 400 may be executed remotely, with fully processed data being transferred to both sides between the processing function 150, for example only for local processing, and the storage function 400, for example for remote storage 400 of processed data.

[00511] For example, an example may be given with respect to a sequencing function, such as through a typical next generation sequencer (NGS), where the computing resources 150 are configured to implement the functions required for sequencing genetic material to produce genetic sequenced data, such as reads, that produce at location 100. Said reads, once generated, for example by local NGS, can then be transmitted, for example over a cloud network 30, for example to be stored 400 at a remote location 300 such that they can be re-fetched from the cloud 30 if necessary, for example for further processing, for example, to perform one or more of the secondary and/or tertiary processing functions, which are performed at a location remote from the storage means 400, i.e. locally. In such an example, local storage resources 150 are intended only to serve as a cache of storage in which data is located while awaiting transfer from or to cloud 30, for example to form remote storage facilities 400.

[00512] Similarly, when the computing function is significant, such as where it requires one or more cores 300 of remote computer clusters to process data, and where the storage requirements for storing the processed data 200 are relatively small compared to the computing resources 300 required for processing data, the data to be processed may be sent, for example over the cloud 30, to be processed by remote computing resources, which may include one or more cores or one or more clusters of computing resources, such as one or more super computing resources. In such an example, after data has been processed by the cloud-based computer core 300, the processed data can then be transmitted over the cloud network 30 for local storage 200 and be made available for use by local computing resources 150, such as for local analysis and/or local diagnostics. .

[00513] An example is a typical secondary processing function, for example when accessing pre-processed sequenced data, i.e. data stored locally 200, for example by local computing resources 100, and transmitted over the Internet cloud network 30 to remote computing facilities 300 for further processing therein, for example when performing a secondary processing function, to obtain Russian processed data, which can then be sent back to local funds 100 to store 200 in them. This may be the case where a local technician generates sequenced read data using local data generation resources 100, such as an automated sequencer, and then sends the data over network 30 to remote computing facilities 300, which then perform one or more functions on said data, such as a Burrows-Wheeler transform or a Needleman-Wunsch and/or Smith-Waterman alignment function over specified sequence data, to generate resultant data that can be transmitted over the Internet 30 to local computing resources 100 for examination by one or more locally administered protocols processing and/or for their local 200 storage.

[00514] However, seamless integration is required between the interaction between local 100 and remote 300 computer processing, as well as between local 200 and remote 400 storage, such as in the hybrid cloud-based system 50 presented herein. In such an example, the system may be configured such that the local 100 and remote 300 computing resources are configured to operate together in a seamless manner such that data to be processed by them can be assigned in real time to the local 200 or remote 300 computing resources without significant degradations due to transmission speed and/or significant degradations in operating efficiency. This may be the case, for example, when the software and/or hardware intended to be deployed or otherwise run on the computing resources are configured to correspond with each other and/or are the same or functionally similar, such as the hardware and/or software the software is implemented in the same way to execute the same algorithms in the same way on the generated and/or received data.

[00515] For example, as can be seen in relation to FIG. 8-9, local computing resources 100 may be configured to generate data and, therefore, may include a data generation engine 110, for example to generate and/or analyze raw data, for example to obtain a sequence file in FASTQ format. Said data generation engine 110 may be a local computer, as described herein, having a processor that may be configured to execute one or more software applications and/or may be hardwired to execute one or more algorithms, such as in a mounting configuration , over the generated and/or received data. For example, the data generation engine 110 may be configured to generate data, such as sequencing data 111, which may be measured data 111a, such as data that can be detected as a change in voltage, ion concentration, electromagnetic radiation, and the like; and/or the data generation engine 110 may be configured to generate and/or process a signal, such as analog or digital signal data, such as data representing one or more nucleotide values in a sequence or chain of related nucleotides. In such an example, the data generation engine 110, such as sequencer 111, may also be configured to pre-process the generated data to perform one or more base determination operations 111c, such as on the data to obtain sequence identification data, such as a file in FASTQ format.

[00516] It should be noted that in this example, the data 111 generated in this manner may be generated locally, for example, by means of data generation and/or computing resources 150, such as an on-chip sequencer; or they can be received remotely, for example by remote computing and/or generation resources, such as remote NGS 300, but transmitted over the cloud 30/50 to local computing resources 100, for example for secondary processing 150 and/or storage thereof in local resources 200 storage, for example, while awaiting further local processing 150. In such an example, when the data generating resources 300 are remote from the local processing 100 resources and/or storage resources 200, the corresponding resources can be implemented such that the remote and/or local protocols storage, remote and local processing and/or exchange of information used by each resource may be capable of seamless and/or seamless integration with each other, for example by running the same, similar and/or equivalent software and/or due to having the same, similar and/or equivalent hardware configurations, and/or due to the use of the same communication and/or transmission protocols, which, within certain acceptable limits, may be implemented during production or at a later date.

[00517] In particular, these functions may be implemented in a hard-wired configuration, for example in which the sequencing function and the secondary processing function are performed on the same or related chips or sets of chips, for example in those in which the sequencer and the secondary processor are directly coupled on the chip, as described herein, or may be implemented through software optimized to allow two remote devices to communicate with each other in a seamless manner. A combination of optimized hardware and software implementations may also be used to perform these functions, as described herein. In various embodiments, data generation resources such as sequencer 111, implemented as software and/or hardware or a combination thereof, may also comprise an initial layer of processors 500 such as a manager, various analytical devices, comparators, logging devices, issuing devices, and the like to assist a data generator 111, such as a sequencer, in converting biological information into raw read data, such as files in FASTQ format 111d.

[00518] This may be done in a similar manner for the execution of other functions that may be deployed on local 100 and/or remote 300 computing resources. For example, local computing resources 100 may include software and/or hardware configured to implement one or more secondary layer processing functions 112-115 700 on data generated remotely and/or locally, such as genetic sequence data, such that processing and its results can be seamlessly shared and/or stored with each other. In particular, the local computing function 100 and/or the remote computing function 300 may be configured to generate and/or receive raw data, such as genetic sequence data, such as a file in FASTQ format or the like, and execute one or more protocols 600 for secondary processing of the generated and/or received data, which protocols may be implemented in software, hardware, or a combination thereof. For example, the data generation and/or computation resources 110 may be configured to implement one or more mapping operations 112, alignment operations 113, or other related functions 114 on the received or generated data.

[00519] More specifically, the data generation resources 110 may comprise a mapping module 112, as described herein, or may otherwise be programmed to perform a mapping algorithm on genetic sequence data, such as to perform a Burrows-Wheeler transform and/or other algorithms constructing a hash table and/or executing a hash function 112a on said data, such as mapping seeds using a hash function to generate mapped sequence data. The data generation resources 110 may also include an alignment module 113, as described herein, or may otherwise be programmed to perform an alignment algorithm on genetic sequence data, such as mapped sequenced data, such as to perform a Smith-Waterman alignment with gaps and/or or without gaps, and/or the Needleman-Wunsch algorithm or other evaluation algorithms 113a on the specified data to generate aligned sequence data. The data generation resources 110 may also be configured to include one or more other modules 114 configured to perform one or more other processing functions on genetic sequence data, such as mapped and/or aligned sequence data, and may therefore suitably contain configured module 114 or may be otherwise programmed to perform one or more other processing functions, such as sort 114a, de-duplicate 114b, recalibrate 114c, local realignment 114d, duplicate marking 114f, Recalibrate base quality evaluation function(s) 114g, and/or or a compression function (e.g., to produce a file in BAM format, Reduced BAM, and/or compressed and/or decompressed CRAM) 114e, in accordance with methods disclosed herein, which processing functions may be implemented as one or more system pipelines 1. Likewise, system 1 may include a module 115 configured to process data, such as sequenced, mapped, aligned and/or sorted data, so as to obtain a variant definition file 116, such as with processing functionality based on hardware and/or or software. More specifically, system 1 may include a variant determination module 115 for executing one or more variant determination functions, such as a hidden Markov model (HMM) function 115a and/or genomic analysis software (GATK), for example in a mounting configuration and/or through one or more software applications, locally or remotely, and/or converter 115b for the same purposes.

[00520] In some embodiments shown in FIGS. 8 and 10, system 1 may include a local computing function 100 that may be configured to use computer processing resources 150 to perform one or more additional computer processing functions on data generated by the system generator 110 or received by the system acquisition engine 120 (as described below), for example by transmission to them, for example by third parties 121, for example, via cloud 30 or hybrid cloud network 50. For example, third party analyzer 121 can deploy remote computing resources 300 to generate relevant data with the need for further processing, such as data genetic sequence or the like, wherein said data can be transferred to system 1 via a 30/50 network for further processing. This may be useful, for example, when the remote computing resources 300 are NGS configured to receive raw biological data and convert it into digital form, such as the form of one or more FASTQ files containing genetic sequence data reads, and when required further processing, for example to determine how the generated object sequence differs from one or more reference sequences, as described herein, and/or the results are required to be subjected to further, for example tertiary, processing.

[00521] In such an example, system 1 may be configured to provide access to one or more persons, such as first, second, and/or third party users, to associated local processing resources 100 and/or suitably configured remote processing resources 300. associated therewith, so as to enable the user to perform one or more quantitative and/or qualitative processing functions 152 on the generated and/or received data. For example, in one configuration, system 1 may include, for example, in addition to pipelines for primary 600 and/or secondary 600 processing, a third layer of processing modules 700 that may be configured to implement one or more processing functions on generated and/or received primary and/or secondary processed data.

[00522] Specifically, in one embodiment, system 1 may be configured to generate and/or receive processed genetic sequence data 111 that has been mapped remotely or locally to generate variant definition file 116, which can then be further processed , for example in system 1, for example in response to an analytical request 121 from second and/or third parties. More specifically, the system 1 may be configured to receive requests for processing from third parties 121, as well as to perform such requested tertiary processing 700 on the generated and/or received data. In particular, system 1 may be configured to generate and/or obtain genetic sequence data 111, obtain said genetic sequence data and perform mapping 112, alignment 113, and/or sorting 114a thereof to produce one or more variant definition files 116 ( VCFs), and further the system 1 may be configured to implement a tertiary processing function 700 on the data, for example with respect to one or more VCF files. System 1 may be configured to implement any form of tertiary processing 700 on the generated and/or received data, such as subjecting it to pipeline processing functions 700, such as to generate genome data 122a, epigenome data 122b, metagenome data 122c, and the like, including co-genotyping 122d, GATK 122e and/or MuTect2 analysis pipelines 122f. In addition, system 1 may be configured to implement an additional layer of processing on the generated and/or processed data, including, for example, one or more of non-invasive prenatal testing (NIPT) 123a, new patient intensive care unit (N/P ICU) 123b, cancer-related diagnostics and/or therapeutic methods 123c conducted in laboratory research (LDT) 123d, applications 123e related to agrobiology (Ag Bio), or other such processing function related 123f to the healthcare field.

[00523] Consequently, in various implementations where the primary user can access and/or configure system 1 or its components directly, such as by directly accessing them, such as through local computing resources 100, as described herein , system 1 may also be configured to be accessible by secondary parties, such as those connected to system 1 via a local network or internal connection 10 to configure and run system 1 in a local environment. In addition, in some embodiments shown in FIGS. 2B, the system may also be capable of being accessed and/or configured by third parties 121, such as through an associated hybrid cloud network 50 connecting third parties 121 to system 1, such as through an application programming interface (API) accessible through one or more graphical user interface (GUI) components. Such a GUI may be configured to provide third parties with access to the system 1 and, using an API, to configure various system components, modules, associated pipelines, and other associated functional elements to generate and/or process data to run only those system components that which are necessary and/or useful to third parties and/or the launch of which is necessary or required.

[00524] Accordingly, in various examples, the system 1 presented herein may be configured by a primary, secondary, or tertiary user of the system. In such an example, system 1 may be configured by a user, and thus its components can be executed in such a way as to enable execution of some, all, or selected analytical resources of the system, such as 152, on data generated, received by the system, or other transmitted to it, for example by a primary, secondary or tertiary user, such that system 1 runs only those parts of the system that are necessary or useful to perform the analytical functions requested by the user to obtain the desired results. For example, for these and other purposes, an API may be contained in system 1, wherein the API contains or is otherwise operationally linked to a graphical user interface (GUI), including an operating menu and/or a corresponding list of system operating options from which the user can select and/or other configure and manage the system and its components as required.

[00525] In such an example, menus and/or GUI system operating options may control one or more user-selectable first-level operations 600, including: sequencing 111, mapping 112, alignment 113, sorting 114a, variant determination 115, and/or other related functions 114 in accordance with the information provided herein, for example in relation to the primary and/or secondary processing functions described herein. In addition, if desired, menus and/or system operation options in the GUI may control one or more second-level operations 700, including: genome pipeline 122a, epigenome pipeline 122b, metagenome pipeline 122c, co-genotyping pipeline 122d, GATK analysis pipelines 122e and/or MuTect2 122f. In addition, menus and system operation options in the GUI may, if desired, control one or more user-selectable third-level operations 800, including: non-invasive prenatal testing (NIPT) 123a, N/P ICU 123b, cancer-related diagnostics and/or therapeutic modalities. 123c laboratory-based research (LDT) 123d, applications 123e related to agrobiology (Ag Bio), or other such processing functions related 123f to the healthcare field.

[00526] Accordingly, menus and system operation options may include one or more of primary, secondary, and/or tertiary processing functions to configure the system and/or its components to implement one or more data analysis pipelines selected and configured by the user. In such an example, local computing resources 100 may be configured to correspond to and/or mirror remote computing resources, and/or similarly, local storage resources 200 may be configured to correspond to and/or mirror remote storage resources 400 , so that various components of the system can be run and/or the data they generate can be stored either locally or remotely in a seamless manner at the discretion of the user of system 1. Additionally, in some embodiments, system 1 can be accessed by third parties to run their own protocols 121a analysis on the generated and/or processed data, for example by running through an artificial intelligence interface, configured to search for correlations between them.

[00527] System 1 may be configured to implement any form of tertiary processing on the generated and/or received data. Therefore, in various embodiments, a primary, secondary, or tertiary user may access and/or configure any layer of the system 1 and its various components directly, such as through direct access to computing resources 100, indirectly, such as through a local network connection 10, or over an associated hybrid cloud network 50 connecting the specified person to system 1, for example through a suitably configured API having appropriate permissions. In such an example, the components of the system may be represented as a menu, such as a selectable GUI menu, in which the user can select from all the different processing and storage options to be performed on data submitted by the user. Additionally, in some examples, the user may load custom system protocols to be accepted and executed by the system to process various data in a manner designed and chosen by the user. In such an example, the GUI and associated API would provide the user with access to System 1 and, using the API, the ability to add and configure various system components, modules, associated pipelines, and other associated functional elements to generate and/or process data to run only those system components. which are necessary and/or useful to the specified person and/or the launch of which is necessary or required.

[00528] Although the above in relation to FIG. 8 and 9, information is directed to data generation 110, such as local data generation 100 using local computing resources 150; as described above in relation to FIG. 9, one or more of the above individual modules and their respective functions and/or associated resources may be capable of being remotely implemented, for example by remote computing resources 300, and may also be capable of being communicated to system 1, for example via a protocol seamless transmission over a 30/50 Internet connection on a cloud basis, for example through a suitably configured data acquisition engine 120.

[00529] Accordingly, in such an example, local computing resources 100 may include a data acquisition engine 120, such as configured to transmit and/or receive such acquired data and/or related information. For example, system 1 may include a data acquisition engine 120 configured to enable continuous processing and/or storage of data in a seamless and reliable manner, such as over a cloud or 30/50 hybrid network in which processing functions are distributed locally 100 and/or remotely 300 , and likewise in which one or more results of such processing may be stored locally 200 and/or remotely 400 such that the system seamlessly allocates to which of the local and remote resources a selected task should be directed for processing and/or storage, outside depending on where the specified resources are physically located. Such distributed processing, transmission, and reception may include one or more of sequencing 111, mapping 112, alignment 113, sorting 114a, duplicate labeling 114c, duplicate removal, recalibration 114d, local realignment 114e, Recalibration function(s) 114f, and base quality assessment function(s), and /or compression functions 114g, as well as variant determination functions 116, as described herein. When processed data is stored locally 200 or remotely 400 in any state, it may in the process be made available to local 100 or remote 300 processing resources, such as for further processing before retransmission and/or re-storage.

[00530] In particular, system 1 may be configured to generate and/or receive genetic sequence data 111, may be configured to process said genetic sequence data 111 locally 150, or transmit said data over a suitably configured cloud 30 or hybrid cloud 50 networks, for example to remote processing means for remote processing 300. In addition, after processing the data, the system 1 may be configured to store the processed remotely 400 or transfer it back to local storage 200. Accordingly, the system 1 may be configured to store the processed remotely 400 locally or remotely. generation and/or processing of data, for example where the generation and/or processing steps may form a first layer of primary and/or secondary processing functions 600, which may include one or more of: sequencing 111, mapping 112, alignment 113, and/or sorting 114a for receiving one or more variant definition files (VCFs) 116. Likewise, system 1 may be configured to locally or remotely generate and/or process data, such as where the generation and/or processing steps may form a second layer of tertiary processing functions 700, which may include one or more of generating and/or receiving data according to genome pipeline 122a, epigenome software pipeline 122b, metagenome pipeline 122c, co-genotyping pipeline 122d, GATK analysis pipeline 122e and/or MuTect2 122f. Additionally, system 1 may be configured to generate and/or process data locally or remotely, for example where the generation and/or processing steps may form a third layer of processing functions, as shown in FIG. 11, which may include one or more of generating and/or obtaining data related to and including: non-invasive prenatal testing (NIPT) 123a, N/P ICU 123b, cancer-related diagnostics and/or therapeutic modalities 123c, various conducted in laboratory research (LDT) 123d, applications 123e related to agrobiology (Ag Bio), or other such processing functions related 123f to the healthcare field.

[00531] In some embodiments shown in FIGS. 8, 9, and 10, the system 1 may also be configured to allow one or more persons to access the system and to transmit information to or from associated local processing 100 and/or remote processing resources, as well as to store information locally 200 or remotely 400 in a manner that allows the user to select what information to process and/or store in the system 1. In such an example, the user can not only decide what primary, secondary and/or tertiary processing functions to perform on the generated and/or received data, but as well as how the specified resources can be implemented and/or where the results of such processing will be stored. For example, in one configuration, the user can select whether data will be generated locally or remotely, or a combination of these, whether it will be processed, and if so, which secondary processing modules will process it, and/or which resources execute which of the following. processes, and can further determine whether the generated or received data will be further subjected to tertiary processing, and if so, which modules and/or which levels of tertiary processing 700 will process it, and/or which resources perform which of said processing, and, in a similar way, where the results of the specified processing will be stored for each stage of operations.

[00532] Specifically, in one embodiment, a user may configure system 1 such that generation of genetic sequence data 111 occurs remotely, such as through NGS, and secondary data processing 600 occurs locally. In such an example, the user can then determine which of the secondary processing functions are performed locally 100, for example, by selecting processing functions such as mapping 112, leveling 113, sorting 111, and/or retrieving VCF file 116 from a menu of available processing options. The user can then choose whether to subject locally processed data to tertiary processing, and if so, which modules to activate for further processing of the data, and whether such tertiary processing will occur locally 100 or remotely 300. Likewise, the user can select different options for various levels of tertiary processing options, and where any generated and/or received data should be stored, locally 200 or remotely 400, at any given stage or at any given point in time during the operation.

[00533] More specifically, a primary user may configure the system to receive processing requests from third parties, wherein the third parties may configure the system to perform the requested primary, secondary, and/or tertiary processing on the generated and/or received data. In particular, a user or second or third parties may configure the system 1 to generate and/or obtain genetic sequence data, locally 100 or remotely 200, may configure the system 1 to obtain said genetic sequence data and perform mapping, alignment and/or sorting thereof, locally or remotely, to obtain one or more variant definition files (VCFs), and may further configure system 1 to perform a tertiary processing function on the data, for example, with respect to one or more VCF files, locally or remotely. More specifically, the user or others may configure the system 1 to perform any form of tertiary processing on the generated and/or received data, as well as determine where such processing will occur in the system. Therefore, in various embodiments, a first, second, or third party user 121 may access and/or configure the system 1 and its various components directly, such as through direct access to the local computing function 100, through a local network connection 10, or over an associated hybrid cloud network 50 connecting said entity 121 to system 1, for example through an application program interface (API) accessible through one or more graphical user interface (GUI) components. In such an example, a third party user may access system 1 and use the API to configure various system components, modules, associated pipelines, and other associated functional elements to generate and/or process data to run only those system components that are needed and/or or useful to third parties and/or whose execution is necessary or required, and also designate which computing resources will support the requested processing and where the resulting data will be stored.

[00534] Accordingly, in various examples, system 1 may be configured by a primary, secondary, or tertiary user of the system, who may configure system 1 to configure its components to enable execution of some, all, or selected analytical resources of the system over data that directly generates, causes them to be generated by system 1, or causes them to be transmitted to system 1, for example via a network associated with it, for example through a data acquisition mechanism 120. In this way, the system 1 can be configured to run only those parts of the system that are necessary or useful to perform the analytical function desired and/or requested by the persons making the request. For example, for these and other purposes, an API may be provided that contains an operating menu and/or a corresponding list of system GUI functional options from which the user can select to configure and control the system as required. In addition, in private embodiments, the system 1 may be made available to third parties, such as government regulatory agencies, such as the Federal Drug Administration (FDA) 70b, or provide third parties with the ability to collate, compile and/or access a database of genetic information, extracted or otherwise derived and/or compiled by the system 1 to form an electronic medical record (EMR) database 70a and/or to provide government access to the system and/or oversight of the system, such as for the FDA for Drug Development Evaluation. System 1 may also accumulate, compile and/or annotate data 70c and/or provide other high-level users with access to it.

[00535] Accordingly, in various embodiments, as can be seen in FIG. 13, a hybrid cloud 50 is provided, configured to connect local computing resources 100 and/or storage resources 200 with remote computing resources 300 and/or storage resources 400, such as when the local and remote resources are separated from each other distally, spatially, geographically etc. In such an example, local and distal resources may be configured to communicate with each other so as to share information, such as digital data, in a seamless manner with each other. In particular, local resources may be configured to implement one or more types of data processing, such as prior to transmission over the hybrid network 50, and remote resources may be configured to implement one or more types of further data processing.

[00536] For example, in one particular configuration, the system 1 may be configured such that the generation and/or analysis function 152 is configured to be performed locally 100 via local computing resources, such as to perform a primary and/or secondary processing function to generate and/or or processing genetic sequence data as described herein. Additionally, in various embodiments, the local resources may be configured to implement one or more tertiary data processing functions, such as one or more of genome, exome, and/or epigenome analysis, or cancer, microbiome, and/or other analyzes with DNA/RNA processing. In addition, when such data is intended to be transferred, for example to remote computing resources 300 and/or storage resources 400, the data may be transformed, for example by a suitably configured converter 151, which may be configured to index, transform, compress and/or encrypt the data , for example before transmission over the hybrid network 50.

[00537] In particular examples, such as when generated and processed data is transferred to remote computing resources 300 for further processing, such processing may be global in nature and may include receiving data from multiple local computing resources 100, collating such data sets, annotating the data, and comparison, for example to interpret data, determine trends, analyze trends of various biomarkers and contribute to the development of diagnostic methods, therapeutic methods and/or prevention methods. Accordingly, in various examples, the remote computing resources 300 may be implemented as a data processing core, such as where data from a variety of different sources may be transmitted, processed, and/or stored, such as by being accessed by the local computing resources 100. More specifically, the remote core 300 processing may be configured to receive data from a plurality of resources 100, process the data, and distribute the processed data back to a plurality of different local resources 100 to enable interaction between researchers and/or resources 100. Such interaction may include various data sharing protocols and may further include preparing the data to be transmitted, for example by allowing the user of the system 1 to choose between different security protocols and/or privacy settings to control how the data is prepared for transmission.

[00538] In one specific example, as shown in FIG. 11, local computing resources 100 and/or storage resources 200 are provided, such as at the user's location. Computing resources 100 and storage resources 200 may be coupled to data generation resources 121, such as NGS and/or an on-chip sequencer as described herein, for example via a direct or internal connection 10, wherein the sequencer 121 is configured to generate genetic data sequences, for example a file in FASTQ format. For example, the sequencer 121 may be part of and/or may be located in the same device as the sequencer of the computing resources 100 and/or the storage unit 200 in order to have a direct connection with the ability to exchange information and/or control, or the sequencer 121 and the computing resources 100 and/or storage resources 200 may be part of separate devices, but may be co-located and thus may be connected via a wired or internal 10 connection. In some examples, sequencer 121 may be located in a separate facility from the facility with computing resources 100 and/or storage resources 200, and thus may be connected over the Internet 30 or hybrid cloud connection 50.

[00539] In such examples, genetic sequence data may be processed 100 and stored locally 200 before being transmitted through a suitably configured converter 151, or the generated sequence data may be transmitted directly to one or more converters 151 and/or analyzers 152, such as through suitable the configured local connection 10, internal 30, or hybrid cloud connection 50 as described above, for example, before performing processing locally. In particular, like the data generation resources 121, the converter 151 and/or the analyzer 152 may be part of and/or located in the same device as the same elements of the computing resources 100 and/or storage unit 200 in order to have a direct connection with it with the ability to exchange information and/or control, or the converter 151 and/or analyzer 152 and computing resources 100 and/or storage resources 200 may be part of separate devices from each other, but may be located in the same facility and, thus, can be connected via a wired or internal 10 connection. In some examples, the converter 151 and/or analyzer 152 may be located in a separate facility from the facility with the computing resources 100 and/or storage resources 200, and thus may be connected over the Internet 30 or hybrid cloud connection 50.

[00540] In examples such as those shown in FIGS. 13, converter 151 may be configured to prepare data for transmission before or after analysis, such as through suitably configured computing resources 100 and/or analyzer 152. For example, analyzer 152 may perform secondary and/or tertiary processing functions on the data. , as described herein, for example, to analyze the generated sequence data with respect to determining their genomic and/or exome characteristics 152a, their epigenomic features 152b, any various DNA and/or RNA markers of interest and/or cancer indicators 152c, and their relationships with one or more microbiomes 152d, as well as one or more secondary and/or tertiary processes, as described herein. As indicated, the generated and/or processed data may be converted, for example through a suitably configured converter 151, for example before being transmitted through the system 1 from one component thereof to another, for example over a direct, local 10, Internet 30 or hybrid cloud 50 connection. Such a transformation may include one or more of a transformation 151d, such that the data is transformed from one form to another; providing understanding 151c, including encoding, decoding and/or otherwise extracting data from an incomprehensible form, and converting it into an understandable form, or from one understandable form to another; indexing 151b, for example including compiling and/or collating generated data from one or more resources and allowing it to be searched and/or located, for example through generated indexes; and/or encryption 151a, such as creating a locked and unlocked password-protected data set, such as before transmission over the Internet 30 and/or hybrid cloud 50.

[00541] Therefore, in these and/or other such examples, the hybrid cloud 50 may be configured to enable seamless and secure transmission of data across all components of the system, for example, when the hybrid cloud 50 is configured to allow different users of the system to configure its component parts and/or the system itself to meet the user's needs for research, diagnostic, therapeutic and/or preventative discovery and/or development. In particular, the hybrid cloud 50 and/or the various components of the system 1 can be configured to be operably connected to compatible and/or appropriate APIs that are configured to allow the user to remotely configure the various components of the system 1 to deploy the required resources in a desired manner, wherein local, remote or a combination of these, for example based on the needs of the system and the specific types of analysis being performed, while ensuring the exchange of information in a secure environment with the ability to encrypt. Another exemplary implementation of the hybrid cloud system presented herein is shown in FIG. 12.

[00542] For example, as shown in FIG. 10-12 and in particular in FIGS. 15, system 1 may be a multi-tiered and/or multiplexed bioanalytical processing platform that contains layers of processing units, each of which has one or more processing pipelines that can be deployed systematically, simultaneously and/or sequentially to process genetic information, starting from the stage 400/500 of its primary processing to obtain genetic sequence data, for example in the form of one or more files in FASTQ format; to a secondary processing step 600 to obtain one or more variant definition files; and further to selecting one or more variant definition files or other associated processed data and performing one or more other operations on them, for example to perform one or more of diagnostic, prophylactic and/or therapeutic procedures on them, for example in response to a request 121 third persons and/or in response to data submitted by third parties 121. Such further processing may include various pipeline protocols 700, such as those configured to perform analytical functions on certain genetic variation data of one or more subjects, including genome, epigenome, metagenome and/or analytics. or genotyping, for example at one level, and/or various types of disease diagnostics and/or research protocols 800, which may include one or more of NIPT, NICU, cancer, LDT, biological, agrobiological applications and the like. In particular, the system 1 may also be configured to receive and/or transmit various data 900 related to the procedures and processes described herein, for example, related to electronic medical records (EMR) data, testing and/or structuring data Federal Drug Administration (USA) related to annotation data and the like. Such data may be useful in allowing the user to create and/or access generated medical, diagnostic, therapeutic and/or preventive methods developed through the use of the system 1 and/or available through it.

[00543] Consequently, one or more, for example all, of these functions can thus be performed locally, for example on site 10, in the cloud 30, or through controlled access through a hybrid cloud 50. In such an example, a development environment is created that allows the user to control the functionality of the system to suit his or her individual needs and/or to make it available to others who require the same or similar results. Consequently, the various components, processes, procedures, tools, levels and hierarchies of the system can be made configurable through a GUI interface that allows the user to select which system components to use, on what data, at what time and in what order according to established user requirements and protocols to generate appropriate data and connections between data that can be securely transmitted throughout the system, locally or remotely. As stated, these components can be configured to seamlessly communicate with each other regardless of location and/or connection type, for example by being configurable to perform the same or similar processes in the same or similar manner, for example by using appropriate APIs , distributed throughout the system, the use of which allows different users to configure different components to perform different procedures in the same way.

[00544] For example, an API may be defined in a header file regarding the processes to be executed by each specific component of the system 1, the header describing the functionality and defining how to call the functions, such as parameters that are passed, inputs received, and outputs passed, and the way in which it happens, what comes in and how, what is issued and how, what is returned and how. For example, in various embodiments, one or more components and/or elements thereof, which may form one or more pipelines of one or more layers of the system, may be configured, such as by instructions entered by a user and/or one or more applications of second and/or third parties. These instructions can be passed to the system through appropriate APIs that communicate with one or more different system drivers, indicating to the driver(s) which parts of the system, such as which modules and/or which processes within them, should be activated, when, and in what order. , given a preselected configuration of parameters, which can be defined using an interface that can be selected by the user, such as a GUI.

[00545] As described above, one or more DMA drivers of system 1 may be configured to operate accordingly, for example, at the core level of each component and system 1 as a whole. In such an example, one or more of the provided cores may have its own very low level core API that provides access to the hardware and functions of various components of the system 1, so that access to the corresponding registries and modules is provided for configuring and managing processes and how they are executed in system 1. Specifically, a virtual layer of utility functions can be formed on top of this layer to form the building blocks that are used for a variety of functions that send files below to the kernel(s) and receive back results, encoding, encrypting and/or transmit the corresponding data and then perform higher-level functions on it. On top of this layer, an additional layer may be formed using said utility functions, which may be an API layer with which the user can interact, and this layer may be configured to function primarily to configure the system 1 as a whole or its component parts by downloading files and downloading results, where files and/or results can be transferred throughout the system, locally or globally.

[00546] Such configuration may involve exchanging information with registries as well as making function calls. For example, as described herein above, one or more function calls necessary and/or useful to perform steps, for example, sequentially, to perform mapping, alignment, sorting and/or variant determination, or other secondary and/or tertiary function, as described herein, may be implemented in accordance with hardware operations and/or corresponding algorithms to generate the necessary processes and perform the required steps.

[00547] In particular, since in certain embodiments one or more of these operations may be based on one or more structures, it may be necessary to construct various structures necessary to perform these operations. This will require a call to a function that performs these actions, wherein said function call will build the required structure to perform the operation, and so said call will take the name of the file where the structure parameter files are stored and then generate one or more data files , which contain and/or configure the required structure. Another call to the function may be to load a structure that has been generated by the appropriate algorithm and transfer it to on-chip memory and/or system 1 and/or place it in the desired location where the hardware expects it to be. Of course, various data will need to be downloaded to the chip and/or otherwise transferred to the system oscillator, as well as various other selected functions of System 1, and these functions can be performed by the configuration manager, for example by downloading everything that is necessary for the pipeline modules platform layers of the chip and/or system as a whole performed their functions in memory located on the chip and/or system or attached to or otherwise associated with the chip and/or system.

[00548] Additionally, as shown in FIG. 16, the API may be configured to allow one or more chips of system 1 to communicate with a circuit board of sequencer 121, compute resource 100/300, converter 151, analyzer 152, interpreter 310, collaborator 320, or other system component when included in thereto, to receive a FASTQ file and/or other generated and/or processed genetic sequence files directly from the sequencer or other processing component, for example immediately after they are generated and/or processed, and then transmit this information to the configuration manager, which then forwards this information into the appropriate memory banks in the hardware and/or software, which make this information available to the appropriate modules of the hardware, software and/or the system as a whole so that they can perform their assigned functions on this information for base determination, mapping, alignment, sorting, etc. DNA/RNA sample relative to the reference genome and/or the execution of associated secondary and/or tertiary processing operations on it.

[00549] Accordingly, various implementations may include a client level interface (CLI) that may allow users to directly access one or more of these functions. In various embodiments, the CLI may be a software application, such as having a GUI, that is configured to configure the availability and/or use of system hardware applications and/or various other software. Thus, a CLI can be a program that accepts instructions, such as arguments, and makes functionality available simply by calling the application program. As stated above, the CLI can be command line or GUI (Graphical User Interface) based. The command level for the command line is below the GUI level, where the GUI contains a window-based file manager having clickable function blocks that depict which modules, which pipelines, which layers of which platforms will be used, as well as the parameters for their use . For example, at runtime, if directed, the CLI will find the reference, determine whether a hash table and/or index needs to be generated, or, if they have already been generated, establish where they are stored, and manage the loading of the generated hash table and/ or index, etc. Instructions of these types may appear as user options on the GUI, which the user can select for execution by the associated chip(s) and/or system 1.

[00550] Additionally, a library may be included that may contain pre-existing editable configuration files, such as files targeting typical user-selectable hardware and/or associated software functionality, such as those related to analysis of a portion or entire genome and/or protein, for example for various types of analysis, such as the analysis of personal medical histories and pedigrees, or the diagnosis of diseases, or the discovery of new drugs, therapeutics and/or one or more other analytical methods, etc. Parameters of these types can be preset, for example to perform these types of analyses, and can be saved in a library. For example, if the platform described herein is used for, for example, NIPT, NICU, cancer, LDT, AgBio, and related research at the aggregate level, the present parameters may be configured differently than if the platform were simply aimed to conduct genomic and/or genealogical research, for example at the individual level. See, for example, FIG. eleven.

[00551] More specifically, in the case of certain diagnostics of an individual, accuracy may be an important factor and thus system parameters may be set to provide increased accuracy, albeit at the expense of a possible reduction in speed. However, in other genome-related applications, speed may be a major determining factor, and thus the system parameters may be set to provide maximum speed, but this may result in some reduction in accuracy. Accordingly, in various embodiments, frequently used parameter settings for performing various tasks may be preset in the library to facilitate their use. Such parameter settings may also include necessary software applications and/or hardware configurations used in the operation of the system 1. For example, the library may contain code that executes the API, and may also contain image files, scripts, and any other supporting information, necessary for the operation of system 1. Consequently, the library can be configured to compile software for executing the API, as well as various executable objects.

[00552] Additionally, as shown in FIG. 12-14, the system may be configured such that one or more system components may be executed remotely, such as when the system component is configured to perform one or more comparison functions on data, such as interpretation function 310 and/or collaboration function 320. For example, when an interpretation protocol is applied to the data, the interpretation protocol 312 may be configured to analyze and draw conclusions about the data and/or determine various relationships regarding it, and one or more other analytical protocols may also be performed, including annotating the data 311, performing diagnostics 313 on the data and/or analyzing the data to determine the presence or absence of one or more biomarkers 314. Additionally, when performing a collaboration protocol, the system 1 may be configured to provide an electronic forum where data can be shared 321, wherein the data sharing protocol may include security settings 324 and/or privacy 322 that can be selected by the user and provide the ability to encrypt data and/or password protect it so that identifiers and data sources can be hidden from the user of system 1. In specific examples, system 1 may be configured to provide third party analyzer 121 the ability to perform virtual simulations on data. In addition, once generated, data interpreted and/or subjected to one or more collaborative analyzes may be stored remotely 400 or locally 200 to make it available to remote 300 or local 100 computing resources, such as for further processing and/or analysis. .

[00553] Accordingly, in view of the disclosure herein, in one aspect, an apparatus is provided for performing one or more of a plurality of operations in performing a genome sequence analysis operation. In some examples, the device may be a "bench" computing solution having a chipset that is coupled to a PCIe card, which itself may be inserted into the computing device to thereby be coupled to one or more internal central processing units (CPUs). CPUs), graphics processing units (GPUs) and/or associated storage devices. In particular, a computing device, processing units, associated storage devices, and/or associated PCIe card(s) having one or more FPGA/ISCH chipsets as described herein may be capable of communicating with each other. other and can be located inside a housing, for example in a boxed version, which is typical in the art and can be configured for use as a workstation and/or it can be provided and/or configured for use in a server rack. In other embodiments, chipsets and/or associated backplane(s) may be interconnected within a next-generation sequencing device to form a single unit therewith.

[00554] However, in various examples, one or more integrated circuits disclosed herein may be provided and accessible through a cloud-based interface, such as as shown in FIG. 12-15. In some examples, the boxed version may be configured to be remotely accessible, such as when the boxed configuration is made portable to the cloud. However, in other cases, one or more of the integrated circuits disclosed herein may be part of a server rack, such as when a system accessible on the server is capable of being accessed remotely, such as through the cloud 50.

[00555] For example, in one embodiment, a server having one or more, eg multiple, CPU and/or GPU cores and associated storage devices may be associated with one or more FPGAs/ISSNs disclosed herein. Specifically, in one embodiment, a boxed version/server may be provided containing 18 to 24 or more CPU cores, having solid state drives (SSDs), 128x8 RAM, and one or more BioIT systems FPGA/ISSN, to which remote access is provided. In one example, one or more FPGAs may be configured to change their configuration, such as partially change their configuration, between one or more different stages of the genomic analysis pipeline. In other examples, the server system may contain up to 36 CPU/GPU cores and approximately 972 GB of RAM, which may be associated with approximately 8 FPGAs configured as described herein.

[00556] More specifically, the proposed FPGAs may be configured to specialize in performing one or more computationally intensive operations in a BioIT pipeline, such as where one FPGA is provided that is specifically designed to perform a mapping operation, and another FPGA is provided that is configured to implement the operation alignment, although, in some examples, there may be one FPGA that is configured to be at least partially reconfigured between the execution of both mapping and alignment operations. Other operations in the pipeline that may be performed by dedicated FPGAs may include performing an HMM operation, a local realignment operation such as a Smith-Waterman operation, and/or various other variant determination operations. Likewise, various pipeline operations may be executed by one or more of the system's associated CPU/GPUs. Such operations may be one or more computationally intensive pipeline operations, such as to perform sorting, deduplication, and other variant determination operations. Therefore, the system may be configured to perform a combination of operations, partly by the CPU and partly by hardware such as the FPGA/ISCH system.

[00557] Accordingly, various system embodiments may provide examples of various CPUs and hard-wired integrated circuits to perform the specialized pipeline analysis functions of genomes proposed herein. For example, there may be different FPGA instances for performing specialized genomic analysis operations, such as an FPGA instance for performing mapping, another instance for performing alignment, another for performing local realignment and/or other Smith-Waterman operations, another for performing HMM operations, and things like that. Likewise, various CPU/GPU instances may be provided to perform specialized genomic analysis operations, such as CPU/GPU instances to perform sorting, deduplication, compression, various variant detection operations, and the like. In such examples, one or more interconnected storage devices may be provided, eg between different computational stages of the pipeline, to receive received data resulting from computation, compilation and processing throughout the system, eg between different instances of the CPU and/or FPGA. In addition, it should be noted that the size of the various CPU and/or FPGA instances may vary depending on the computing needs of the system and may vary from small to medium, large and very large, similarly the number of CPU/GPU and FPGA instances may vary. ISSN.

[00558] Consequently, the system may also include a workflow manager that is configured to schedule and direct the movement of data throughout the system and from one instance to another or from one memory to another. In some examples, the storage device may be a dedicated instance-specific storage device, and in other examples, the storage device may be configured to be elastic and thus able to switch from one instance to another, such as an elastic block storage storage device . In other examples, the storage device may be non-instance specific and therefore be configured to communicate with multiple instances, such as for elastic file storage.

[00559] In addition, the worker thread manager may be a specialized instance, such as a CPU/GPU core, that is specifically designed for and/or configured to determine which jobs need to be executed and when and what resources will be used in executing said jobs, and also queuing tasks and directing them from resources to resources. The worker thread manager may comprise or may otherwise be implemented as a load estimator and/or may form an elastic control node, which is a specialized instance that may be executed by a processor, such as a CPU core with four cores, or may be executed without multiple cores. In various examples, the workflow manager may have a database connected to it, which may be configured to manage all of the jobs that are to be processed, are being processed, or have been processed. Therefore, the manager can be configured to detect and manage data flows throughout the system, determine how to assign system resources, and when to bring more resources into use.

[00560] As stated above, in certain examples a solution may be provided for a workstation where the system contains multiple servers with X CPU cores that feed data to an FPGA of size Z, where X, Y and Z are numbers that may vary depending depending on the processing needs of the system, but they should be selected and/or otherwise configured to be optimal. For example, typical system configurations are not optimized to perform the BioIT operations of the system described in this document. In particular, certain system configurations are not optimized to maximize data flow from different CPU/GPU instances to different integrated circuits, such as FPGA, systems. More specifically, the system architecture may be designed such that the CPU/FPGA hardware operates in an optimally efficient manner to keep the platforms of both instances constantly busy during operation.

[00561] Therefore, while it is generally good for an FPGA to have great capabilities, the ability to process a large amount of data may not be effective if there is not enough data being fed into the system to be processed. For example, in various examples, quad-core CPU cores may be configured to feed data to a medium-sized FPGA, such as a 2.5X. However, when the CPU is not generating enough work to keep the FPGA busy and/or fully utilized, the configuration will not be ideal. In such configurations, the CPU instance(s) do not generate enough work to keep the available FPGAs busy. Thus, this document proposes an architecture and a method for implementing it that is designed such that the system operates in such a way that the CPU/FPGA software/hardware operates efficiently to ensure that the available CPUs optimally feed data to the available FPGAs in such a way that ensure that both available platforms are always occupied during operation. Accordingly, making such a system accessible via the cloud will ensure that the amount of data fed into it is queued by the worker thread manager and routed to specific CPU/FPGA resources that are capable and configured to receive and process the data in an optimally efficient manner, e.g. when the CPU processes less computationally intensive data and the FPGA performs computationally intensive tasks, and storage devices provide data storage between different procedure steps and/or between different instance types and instances, thereby eliminating inter-instance latency. In particular, there is very little CPU/GPU usage during mapping and alignment because, due to the computationally intensive nature, these tasks are designed to be performed by hardware implementations. Likewise, during case definition, the specified tasks can be divided so that they are roughly equally distributed among the CPU/FPGA instances as their tasks, for example where HMM and Smith-Waterman operations can be performed by hardware and various other operations can be executed by software running on one or more CPU/GPU instances.

[00562] Accordingly, the architecture parameters specified herein are not necessarily limited to the architecture once established; rather, the system is designed to have greater flexibility in the organization of its implementations and relies on a workflow manager to determine which instances are active, when, how, and how long, and specifies which calculations are performed on which instances. In particular, the cloud-based architectures described herein, such as those shown in the accompanying drawings, demonstrate that various known shortcomings of previous architectures can result in inefficiencies that can be overcome by flexibly allowing more CPU/GPU cores to access different instances of hardware, such as FPGAs, that are organized in a more targeted manner to be able to specifically assign the appropriate instance to perform the intended functions to be optimized through implementation in such a format, so that a large portion of the available CPU instances can be kept busy producing the resulting data , which can be optimally fed to the available FPGA instances to ensure continuous occupancy of the selected FPGA instances. Thus, it is desirable to provide a structured architecture that is as efficient as possible and is always occupied. It should be noted that configurations in which too few CPUs supply too many FPGAs such that one or more FPGAs are underutilized are inefficient and should be avoided.

[00563] In one embodiment, the architecture may be configured to virtually include multiple different layers, such as a first layer having a first number of X CPU cores, such as 4 to about 30 CPU cores, and a second layer having 1 to 12 FPGA copies, and the size can vary from small to medium or large, etc. A third level of CPU cores and/or a fourth level of additional FPGAs, etc. may also be included. Therefore, since there are multiple instances available in the cloud, for example instances that simply contain a CPU or GPU and/or FPGA, or combinations thereof, for example at one or more of the layers described herein. Accordingly, in a similar way, the architecture can be organized so that the most intensive special computing functions are performed by hardware instances, and those functions that can be executed by the CPU are directed to a suitable CPU at a suitable level for general processing purposes.

[00564] For example, the architecture may be designed to maximize efficiency and reduce latency by combining different instances at different virtual layers. In particular, a plurality, such as a significant portion and/or all, of Level 1 CPU instances may be configured to supply different Level 2 FPGA instances, such as F12X, that are specifically configured to perform special functions such as mapping, alignment, Smith operation, etc. Waterman, HMM, identifying options and the like. Therefore, Tier 1 CPUs can be used to form the first layer of the genomic analysis pipeline to perform general processing steps and queue data preparation for pipeline analysis, where the data, after processing by one of multiple CPUs, can be fed to a dedicated FPGA instance at the tier 2, for example F12X, for example where the FPGA instance F12X is configured to perform computationally intensive functions such as mapping and/or alignment functions, etc.

[00565] Thus, in a particular embodiment, the CPU instances in the pipeline send their data, once prepared, to one or two FPGA instances for mapping and alignment. Once mapping has been performed, the resulting data may be stored in a storage device and/or then provided to an alignment instance where alignment may be performed, for example by at least one dedicated layer 2 FPGA instance. Similarly processed mapped and aligned data may then be stored in storage device and/or directed to a level 3 CPU instance for further processing, which may be the same as at level 1 or a different instance, for example to perform a less resource-intensive genomic analysis processing function, for example to perform a sorting function. Additionally, after the Tier 3 CPUs have completed their processing, the resulting data can be routed either back to other Tier 2 FPGA instances or to Tier 4 FPGA instances, for example for resource-intensive further genomic processing functions such as Smith-Waterman processing functions (SW), for example in an FPGA instance specifically designed for SW. Likewise, after SW analysis has been performed, for example by an F12X FPGA dedicated to SW, the processed data can be sent to one or more associated storage devices and/or downstream for processing, for example to a CPU instance and/or FPGA at level 4 or 5, or back to level 1 or 3, for example to perform HMM analysis and/or variant determination, for example in a dedicated FPGA and/or a further level CPU processing core.

[00566] Likewise, latency and efficiency issues can be overcome by combining all sorts of different instances at one or more different layers to provide a pipeline platform for genomic processing. Such a configuration may involve more than scaling and/or combining instances; instances may be designed such that they are specialized to perform specially designated functions. In such a case, the mapping FPGA instance does only the mapping, and similarly, the alignment FPGA instance only does the alignment, etc., instead of one instance doing the processing in the pipeline from start to finish. Although in other configurations one or more FPGAs may be at least partially reconfigured, for example between execution of pipeline tasks.

[00567] Therefore, the pipeline manager can manage a queue of genomic processing requests generated by Tier 1 CPU instances to break them down into discrete jobs, aggregate them, and route them to the appropriate job-specific CPU/F1 for processing, such as mapping and/or alignment. wherein the F1 data, after processing, can be sent back or forth to the next level of CPU/FPGA processing of the resulting data, for example to perform various steps in the variant determination module. For example, the case determination function may be broken down into multiple operations that can be performed in software, then directed to HMM processing in one or more hardware FPGA instances, and then sent to the CPU to continue the case determination operations, such as when all The platform is elastically sized and implemented to minimize the cost of expensive FPGA instances while maximizing utilization, minimizing latency, and therefore optimizing operations. Accordingly, fewer hardware instances are thus required due to their absolute processing capabilities and hardware implementation specificity, and therefore the number of FPGAs relative to the CPU can be minimized and their utilization, such as FPGAs, can be maximized, and therefore the system can be optimized. to keep all instances always busy. This configuration is optimally designed for genomic processing analysis, especially mapping, alignment, and variant detection.

[00568] An additional structural element that can be added, for example as an auxiliary device, to the pipeline architecture described herein is one or more elastic memory modules that can be configured to function to provide block storage of data, such as result data , as they are transferred throughout the conveyor. Accordingly, one or more elastic block data storage (EBDS) modules may be inserted between one or more processing layers, such as between different instances and/or instance layers, such that as data is processed and results are obtained, the processed results can be sent to the device EBDS for storage before forwarding to the next level of processing, for example through a dedicated FPGA processing module. The same EBDS can be used between all instances or levels of instances, or multiple EBDS can be used between different instances and/or levels of instances, for example, to store, compile and/or queue result data.

[00569] In this configuration, before sending data directly from one instance and/or from one processing layer to another, the data may be directed to an EBDS or other storage device or structure for storage and subsequent forwarding to the appropriate hardware processing module. Namely, a block storage module can be attached to a node as a storage device, where data can be written to a block data store (BDS) for storage at one level, and the BDS can be switched to another node to forward the stored data to the next level processing. Thus, one or more, such as multiple, BDS modules may be included in a pipeline and configured to switch from one node to another to participate in the movement of data throughout the pipeline. Additionally, as noted above, a more flexible file storage device may be used, such as a device that is configured to be associated with one or more instances at a time, for example, without having to switch from one to another.

[00570] Accordingly, there are multiple steps in a processing pipeline, such as at its service nodes, in preparing data for processing, such as pre-processing, where once prepared, the data is directed to appropriate processing instances at the same level, where resultant data can be generated, The resulting data can then be stored, for example on an EDS device, queued and prepared for the next stage of processing by switching to the next instance node and forwarding to the next instance for processing using the next order FPGA and/or CPU processing instances where they can further result data can be generated, and again once generated, the result data can be sent back to the same or forward to the next EDS level for storage before advancing to the next processing stage.

[00571] Specifically, in one particular embodiment, the flow through the pipeline may look like this: CPU: data prepared (queued and/or stored); FPGA: Mapping, temporary storage, FPGA: alignment, temporary storage; CPU: sorting, temporary storage, CPU: removal of duplicates, temporary storage; FPGA: HMM, temporary storage, CPU: define 1 options, temporary storage, FPGA: Smith-Waterman, temporary storage, CPU: define 2 options, temporary storage, CPU: VCGF, temporary storage, etc. It should be noted that one or more of these steps may be performed in any logical order and may be implemented by any suitably configured resources, such as implemented in software and/or hardware in a variety of different combinations. In addition, one or more EDS or other suitably configured data and/or file storage devices may be attached to one or more of the different nodes, eg between different levels of instances, eg for temporary storage between multiple different processing steps. Accordingly, in a similar manner, each level of processing instances can be elastically scaled as needed, for example between each of the different nodes or levels of nodes, for example to process one or more genomes.

[00572] In accordance with another aspect, as shown in FIG. 16, a method is provided for using a system to generate one or more data files on which one or more secondary and/or tertiary processing protocols can be performed. For example, the method may include providing a genomic infrastructure, such as for one or more of local, cloud and/or hybrid generation, processing and/or analysis in the field of genome and/or bioinformatics.

[00573] In such a case, the genomic infrastructure may include a bioinformatics processing platform having one or more storage devices that are configured to store one or more configurable processing structures for configuring the system to enable one or more analytical processing functions to be performed on data, such as data containing the genomic sequence of interest or processed result data related thereto. The memory may contain a genomic sequence of interest to be processed, for example, after it has been generated and/or obtained, one or more reference genetic sequences and/or may further comprise an index of the one or more reference genetic sequences and/or a list of boundaries relating thereto splicing. The system may also include an input device having a platform application programming interface (API) for selecting from a list of options one or more configurable processing structures, e.g., to configure the system, e.g., by selecting system processing functions to be executed on the data, e.g. pre- or post-processing of genomic sequences of interest. It is also possible to have a graphical user interface (GUI) that can be functionally linked to an API, for example to provide a menu through which the user can select which of the available options to perform on the data.

[00574] The system may be implemented on one or more integrated circuits, which may be formed from one or more sets of configurable, eg preconfigured, or hardwired digital logic circuits, which may be interconnected through a plurality of physical electrical connectors. In such an example, the integrated circuit may have an input, such as a storage device interface, for receiving one or more configurable fabric protocols, such as from a storage device, and may also be configured to implement one or more structures on the integrated circuit according to with configurable processing structure protocols. The input storage interface may also be configured to receive genomic sequence data, which may be represented as a plurality of genomic data reads. The interface may also be configured to access one or more genetic control sequences and index(es).

[00575] In various examples, digital logic circuits may be implemented as a set of processing modules, each formed from a subset of digital logic circuits. Digital logic circuits and/or processing engines may be configured to implement one or more preconfigurable primary, secondary, and/or tertiary processing protocol steps to generate multiple genomic sequence data reads and/or process multiple genomic data reads, such as in accordance with a control genetic sequence(s) or other information derived from a genetic sequence. The integrated circuit may also have an output to output primary, secondary, and/or tertiary processing data, for example, in accordance with a platform application programming interface (API).

[00576] In particular, in various embodiments, digital logic circuits and/or sets of processing modules may form multiple pipelines for genomic processing, such as where each pipeline may have an input that is defined in accordance with the platform application programming interface for receiving the resulting bioinformatics processing platform primary and/or secondary processing data and to perform one or more analytical processes on them in order to obtain the resulting data. In addition, the plurality of genomic processing pipelines may have a common pipeline API that defines a secondary and/or tertiary processing operation to be performed on the resulting primary and/or secondary processing data, such as when each of the plurality of genomic processing pipelines is configured to perform subsets of secondary and/or tertiary processing operations and output of the resulting secondary and/or tertiary processing data in accordance with the pipeline API.

[00577] In such cases, a plurality of genomic analysis applications may be stored in memory and/or associated searchable application storage, such as where each of the plurality of genomic analysis applications may be accessible to a computer via an electronic medium, such as for execution by a computer processor, with for the purpose of performing targeted analysis of genomic data from pre- or post-processing of the resulting primary, secondary and/or tertiary processing data, for example by one or more of a plurality of genomic processing pipelines. In specific examples, each of a variety of genomic analysis applications may be defined by an API and may be configured to receive primary, secondary, and/or tertiary processing resultant data and/or perform targeted analysis of preprocessing or postprocessing genomic data and output the target analysis resultant data to one or more genomic databases.

[00578] The method may further include selecting, for example from a GUI menu, one or more genomic processing pipelines from a plurality of available genomic processing pipelines of the system; selecting one or more genomic analysis applications from a plurality of genomic analysis applications that are stored in the application store; and executing, by a computer processor, one or more selected genomic analysis applications to perform targeted analysis of genomic data from the resulting primary, secondary, and/or tertiary processing data.

[00579] Additionally, in various embodiments, mapping, alignment, sorting, and variant detection may occur on-chip, and in various embodiments, local re-alignment, duplicate marking, base quality assessment recalibration, and/or one or more protocols and/or pipelines for tertiary processing are also performed on-chip, and in various examples, various compression protocols such as BAM and CRAM can also be performed on-chip. However, after the data has been created by primary, secondary and/or tertiary processing, it may be compressed, for example before transmission, such as sending it throughout the system, sending it to the cloud, for example, to run a variant determination module, a secondary, tertiary and/or platform other processing, for example including protocol analysis of interpretation and/or collaboration. This can be useful, especially given the fact that identifying variants, including their tertiary processing, can be “shooting at a moving target”, for example there is no standardized, agreed upon algorithm used in the industry.

[00580] Therefore, different algorithms may be used as needed to achieve different types of results, for example by remote users, and therefore it is useful to have a cloud-based module to perform this function to provide flexibility in choosing the algorithm that is useful at any given specific moment, and also sequential and/or parallel processing. Accordingly, any of the modules described herein may be implemented either in hardware, such as on a chip, or in software, such as in the cloud, but in certain embodiments, all modules may be configured to implement their functions on a chip, or all modules may be configured to perform their functions remotely, for example in the cloud, or a combination of modules can be provided such that some are on one or more chips and others are located in the cloud. Additionally, as noted, in various embodiments, the chip itself or the chips themselves may be configured to operate in conjunction with, and in some embodiments, directly interact with, a genetic sequencer such as NGS and/or an on-chip sequencer.

[00581] More specifically, in various embodiments, a device of the present invention may be a chip, such as a chip, that is configured to process genome data, for example, by using pipeline modules for data analysis. Accordingly, as shown in FIG. 17-19, a genomic pipeline processor chip 100 is proposed along with associated hardware of a genomic pipeline processor system 10. The chip 100 has one or more connections to an external storage device 102 ("DDR3 Memory Manager") and a connection 104 (e.g., a PCIe interface) with the outside world, such as, for example, the host computer 106. The switch 108 (eg, a switch) provides access to the storage device interfaces to various requesters. DMA engines 110 transfer data at high speed between the host and external memories 102 of the processor chip 100 (via switch 108) and/or between the host and the central control unit 112. The central control unit 112 controls the operations of the chip, such as coordinating the actions of multiple modules. 13 treatments. The processing engines are formed from a set of hard-wired digital logic circuits that are interconnected by physical electrical connections and organized into clusters of 114 modules. In some implementations, engines in the same cluster share one switch port through an arbiter. The central controller 112 has connections to each of the clusters of modules. Each module cluster 114 has a number of processing modules for processing genomic data, including a mapper 120 (or mapping module), an aligner 122 (or alignment module), and a sorter 124 (or sorting module). The module cluster 114 may also contain other engines or modules.

[00582] According to one data flow model consistent with the embodiments described herein, the host device sends commands and data through the DMA engines 110 to the central control device 112, which distributes the data evenly among the processing modules. The processing engines return the processed data to the central control device 112, which streams it back to the main device through the DMA modules 110. This data flow model is tailored for mapping and alignment.

[00583] According to an alternative data flow model consistent with the embodiments described herein, the host device streams data to an external storage device, either directly through the DMA modules 110 and switch 108, or through a central control device 112. The master device sends commands to the central control device 112, which sends commands to the processing engines telling the processing modules what data to process. The processing engines access input data from the external storage device, process it, and write the results back to the external storage device, reporting status to the central control device 112. The central control device 112 either streams the resulting data back to the main device from the external storage device , or notifies the host device to itself retrieve the resulting data through the DMA modules 110.

[00584] In FIG. 17 and 18 show a genomic pipeline processor system depicting a complete set of processing modules within a cluster of 114/214 modules. The pipelined processor system may comprise one or more module clusters 114/214. In some embodiments, pipeline processor system 20 includes four or more module clusters 114/214. Processing modules or types of processing modules may include, but are not limited to, a mapper, an aligner, a sorter, a local re-aligner, a base quality recalibrator, a duplicate flagger, a variant detector, compression means and/or decompressor means. In some embodiments, each module cluster 114/214 has one of each type of processing engine. Accordingly, all processing engines of the same type can access the switch 108 simultaneously through different switch ports, since each of them resides in different module clusters 114/214. The formation of 114/214 processing modules of each type in each cluster is not required. Types of processing engines that require enormous parallel processing or storage throughput, such as a mapper (with attached aligner(s)) and sorter, may be contained in each cluster of modules of the pipeline processor system 20. Other types of engines may only appear in one or more clusters of 114/214 modules as needed to meet their or the pipelined processor system's performance requirements.

[00585] In FIG. 19 illustrates a genomic pipeline processor system showing, in addition to the clusters of modules described above, one or more embedded central processing units (CPUs) 202. Examples of such embedded CPUs include Snapdragon® cores or standard ARM® cores. These CPUs execute fully programmable bioinformatics algorithms such as advanced variant detection. Such processing is accelerated by computational functions in clusters of modules that can be invoked by CPU cores 202 as needed. Additionally, even engine-centric processing such as mapping or alignment can be performed by the CPU cores 202, providing them with increased programmability.

[00586] In FIG. 20 shows a processing flow for a genomic pipeline processor system and method. In some preferred embodiments, the data is processed in three passes. The first pass involves mapping 402 and alignment 404, with the full set of reads being run through the engines. The second pass includes sorting 406, where one large block to be sorted (eg, a substantial portion of all reads previously mapped to one chromosome) is loaded into storage, sorted by processing modules, and returned to the central computer. The third pass includes the next steps in the chain (local realignment 408, duplicate labeling 410, base quality score recalibration (BQSR) 412, BAM inference 414, reduced BAM inference 416, and/or CRAM compression 418). The steps and functions of the third pass can be performed in any combination or subcombination and in any order in a single pass. A virtual pipeline architecture, such as the one described above, is used to stream reads from the host to circular buffers in storage, through one processing unit after another, sequentially, and back to the host. In some embodiments, CRAM decompression may be a separate streaming function. In some embodiments, BAM output 414, reduced BAM output 416, and/or CRAM compression 418 may be replaced by variant detection, compression, and decompression.

[00587] Various examples describe a hardware implementation of a sequence analysis pipeline. In fig. 21 is a general block diagram of an embodiment of the present invention. Block 1 describes the hardware implementation of the sequence analysis pipeline. This can be done in many different ways, for example using an FPGA, ISSN, or structured ISSN embodiment. Functional blocks that are implemented using FPGA, ISSN or structured ISSN are shown in FIG. 6 and 7. FIG. 6 and 7 include a number of blocks or modules for performing sequence analysis. The input to the hardware implementation may be a file in FASTQ format, but this format is not a limitation. In addition to the FASTQ file, the FPGA, ISSN, or structured ISSN input data contains auxiliary information, such as Flow Space Information, related to a technology such as Ion Torrent. The number of blocks or modules in FIG. 21 may include the following blocks: error correction, mapping, alignment, sorting, local realignment, duplicate labeling, base quality assessment recalibration, BAM and side information reduction, and/or variant identification.

[00588] These blocks or modules may be present internally, or may be implemented in hardware, but to achieve the purpose of implementing a sequence analysis pipeline, some of these blocks may be omitted and other blocks added. Blocks 2 and 3 describe two alternative pipeline platforms for sequence analysis. The sequence analysis pipeline platform comprises a FPGA, ISSN or structured ISSN and software supported by a host device (eg, PC, server, cluster or cloud computing facilities) using cloud and/or cluster storage. Blocks 4-7 describe the various interfaces that a sequence analysis pipeline may have. In blocks 4 and 6, the interface may be a PCIe interface, but is not limited to a PCIe interface. In blocks 5 and 7, the hardware (FPGA, ISSN or structured ISSN) can be directly integrated into the sequencer. Blocks 8 and 9 describe the integration of a hardware pipeline for sequence analysis integrated into a host system such as a PC, server cluster, or sequencer. The FPGA, ISSN, or structured ISSN hardware is surrounded by multiple DDR3 memory elements and a PCIe interface. The FPGA/ISCH/SSCH board is connected to a host computer consisting of a main CPU, which may be a low-power CPU such as an ARM®, Snapdragon®, or any other processor. Block 10 shows an API hardware pipeline for sequence analysis that can be accessed by third party applications to perform tertiary analysis.

[00589] Accordingly, in various embodiments, a device of the present invention may comprise a computing architecture, such as embedded in an FPGA or a special purpose silicon integrated circuit (SPIC) 100, as shown in FIG. 6 and 7. The integrated circuit 100 may be integrated into a printed circuit board (PCB) 104, such as a peripheral component interface express (PCIe) card, that may be inserted into a computing platform. In various examples, as shown in FIG. 6, the PCIe card 104 may contain a single FPGA or ISSN 100, where the integrated circuit may be surrounded by local storage devices 105, however, in different embodiments, as shown in FIG. 7, PCIe card 104 may include a plurality of FPGAs and/or ISSNs 100A, 100B, and 100C. In various examples, the PCI card may also include a PCIe bus. The PCIe card 104 can be added to a computing platform to run algorithms on extremely large data sets. Accordingly, in various examples, the entire genome sequencing workflow involving an integrated circuit may include the following: sample preparation, alignment (including mapping and alignment), variant analysis, biological interpretation, and/or custom applications.

[00590] Therefore, in various embodiments, the device of the present invention may comprise a computing architecture that achieves highly efficient execution of algorithms, such as mapping and alignment algorithms, that operate on extremely large data sets, such as when the data sets exhibit poor locality of reference (LOR) . These algorithms are designed to reconstruct the entire genome from millions of short read sequences from modern so-called next-generation sequencers, requiring multi-gigabyte random access data structures. Once reconstruction is achieved, as described above in this document, further algorithms with similar characteristics are used to compare one genome with libraries of others, perform functional analysis of genes, etc.

[00591] There are two main approaches currently in use, general purpose multi-core CPUs and general purpose graphics processing units (GPPUs). In such an example, each CPU in a multi-core system might have a classic cache-based architecture, where instructions and data are fetched from a level 1 cache (L1 cache), which is small but extremely fast to access. Multiple L1 caches can be connected to a larger but slower L2 cache. The L2 cache can be coupled to a larger but slower DRAM (Dynamic Random Access Memory) memory system, or it can be coupled to an even larger but slower L3 cache, which can then be coupled to DRAM. An advantage of this arrangement may be that applications in which programs and data exhibit reference locality behave almost as if they were running on a computer with a single storage device as large as DRAM but as fast as an L1 cache. Since fully custom, highly optimized CPUs operate at very high clock speeds, such as 2 to 4 GHz, this architecture can be essential to achieving good performance.

[00592] In addition, this architecture can be extended using GPON, for example by implementing a very large number of small CPUs, each with its own small L1 cache, with each CPU executing the same instructions on different subsets of data. This represents the so-called SIMD (single instruction stream, multiple data stream) architecture. Savings can be achieved by sharing instruction fetching and decoding logic across a large number of CPUs. Each cache has access to multiple large external DRAMs through a network of interconnects. Assuming that the computation must be performed in a highly parallel manner, GPUs have a significant advantage over general-purpose CPUs due to the large amount of computational resources they have. However, they still have a cache-based architecture and their performance is degraded by applications that do not have a high enough degree of reference locality. This results in high cache miss rates and idle processors waiting for data to arrive from external DRAM.

[00593] For example, in various examples, dynamic RAM may be used as system memory because it is more cost effective than static RAM (SRAM). Approximately, we can assume that at the same cost, the volume of DRAM is 4 times greater than the volume of SRAM. However, with the demand for SRAM falling in favor of DRAM, the gap between the two has widened significantly due to the space savings offered by DRAMs, which are in high demand. Regardless of cost, DRAM is 4 times denser than SRAM placed on the same silicon area, as it requires only one transistor and one capacitance per bit compared to 4 transistors per bit to implement a flip-flop in SRAM. DRAM represents one bit of information as the presence or absence of charge in the capacitance. The problem with this structure is that the charge weakens over time, so it needs to be renewed periodically. This need has led to architectures that organize storage devices into independent blocks and access mechanisms that issue multiple words of memory per request. This compensates for the time when a given block is unavailable during an update. The idea is to move a huge amount of data while a given block is available. This is in contrast to SRAM, in which any location in the memory device is accessible in one access for a constant amount of time. This characteristic allows you to focus on one word, rather than a block, when accessing a storage device. DRAMs work well in a cached architecture because each cache miss results in a block of memory being read from the DRAM. The theory of locality of references is that immediately after a reference to the word N, there will likely be a reference to the words N + 1, N + 2, N + 3, etc.

[00594] In FIG. 7 shows an alternative embodiment to FIG. 6 having a plurality of chips 100A, 100B, 100C, where each chip may include one or more different genomic and/or bioinformatics processing modules, such as an exemplary analysis pipeline as disclosed herein.

[00595] In FIG. 17 shows a system 100 for running a sequence analysis pipeline on genetic sequence data. System 100 includes a configuration manager 102, which contains the computing system. The configuration manager computing system 102 may comprise a personal computer or other computer workstation, or may be implemented by a set of networked computers. The configuration manager 102 may also contain one or more third party applications connected to the computing system through one or more APIs that, together with one or more proprietary applications, generates a configuration for processing genome data from a sequencer or other genome data source. The configuration manager 102 also contains drivers that load the configuration into the genome pipeline processor system 10. The genome pipeline processing system 10 may output the resulting data to or be accessible through the Internet 504 or other network to store the resulting data in an electronic health record 200 or other knowledge database 400 .

[00596] As noted several times earlier herein, a genome pipeline processor chip may be coupled or integrated with a sequencer. This chip may also be coupled or integrated on an expansion card, such as a PCIe card, and the expansion card may be coupled or integrated with the sequencer. In other embodiments, the chip may be coupled or integrated with a server computer that is coupled to the sequencer to transfer genomic reads from the sequencer to the server. In yet other embodiments, the chip may be coupled to or integrated with a server in a cloud computing cluster of computers and servers. The system may include one or more sequencers connected (eg, via Ethernet) to a server containing an chip, wherein reads of genomes are generated by the plurality of sequencers, transmitted to the server, and then mapped and aligned on the chip.

[00597] For example, typically in next-generation sequencer (NGS) data pipelines, the processing of the primary analysis step is typically specific to a given sequencing technology. This primary analysis step is performed to translate the physical signals detected within the sequencer into nucleotide sequence “reads” along with associated quality (confidence) scores, such as files in FASTQ format or other formats containing sequence and typically quality information . Additionally, once such a format is obtained, secondary analysis as described herein is performed to determine the contents of the sequenced DNA (or RNA, etc.) sample, for example by mapping and alignment of reads to a reference genome, sorting, labeling of duplicates, recalibration of base quality assessment, local realignment and identification of options. Tertiary analysis may then follow to extract medical and research conclusions from specific DNA contents.

[00598] However, the primary analysis, as mentioned above, is often quite specific in nature to the sequencing technology used. In various sequencers, nucleotides are detected by measuring electrical charges, electrical current, or emitted light. Some primary analysis pipelines often include: signal processing to amplify, filter, separate, and measure the sensor output; reducing the volume of data, for example, by dividing into subgroups, thinning, averaging, transforming, etc.; image processing or digital processing to identify and amplify significant signals and associate them with specific reads and nucleotides (eg, image bias calculation, cluster identification); algorithmic processing and heuristic procedure to compensate for sequencing technology artifacts (eg, phasing estimation, crosstalk matrices); Bayesian probability calculations; hidden Markov models; base determination (selection of the most likely nucleotide at each position in the sequence); assessment of the quality (reliability) of determining the grounds, etc.

[00599] The primary analysis can be highly commutatively intensive, sometimes as extensive as the secondary analysis. For example, in existing sequencing technologies, primary analysis often uses FPGA and GPU to speed up the process beyond the CPU's capabilities. But these accelerated functions can be performed much more efficiently in a specific integrated circuit, such as the one described herein. For example, they can be implemented in a structured ISSN using configurable metal layers, since they do not require as much precision in physical placement as embedded processor cores require, however, significant parallel computations implemented in large FPGAs and GPUs may be difficult to implement in the configurable resources of a structured ISSN. Alternatively, it is possible to implement the primary processing acceleration logic in the base plate of the structured ISSN, taking advantage of the spatial efficiency of a standard cell in the base plate.

[00600] The reason that secondary processing functions can be implemented in configurable metal layers of structured ISSN is that algorithms for secondary processing of genome data are still under active research. Thus, it may be an advantage to be able to periodically update the structured ISS design in a low-cost manner, for example every year or two years to apply the newest algorithms. In contrast, the primary infusion analysis algorithms currently in use are more mature, and the required processing steps have been researched and defined by the respective sequencer manufacturers. Although still subject to some modification, these algorithms are more generic signal and numerical processing than secondary analysis, so proper configurability and microprogramming of the primary processing acceleration modules can make them flexible enough for significant changes. With embedded cores, processors enhance this flexibility even further. Therefore, it makes sense to integrate methods for accelerating primary processing into the base plate of FPGA and/or structured ISSN, as described herein.

[00601] It is also advantageous to integrate primary processing acceleration and secondary processing acceleration into a single integrated circuit FPGA or ISSN (standard cell or structured ISSN) with or without embedded processors. This can be beneficial because sequencers produce data that requires primary and secondary analysis, and integrating them into a single device is most efficient in terms of cost, space, energy, and resource sharing. If embedded processors are also present, they can be used to increase the speed and flexibility of both primary and secondary processing.

[00602] These three components—primary accelerators, secondary accelerators, and embedded processors—can be implemented in an FPGA baseplate or a structured ISSN and/or using configurable metal layers, or a combination thereof. All three could be in the base plate, or all three could use configurable metal layers, or any two of them could be in the base plate and the others use configurable metal layers. In any of these configurations, all three can communicate with each other in any combination of indirect and/or through storage, and cooperate in common tasks. One preferred configuration is to implement primary acceleration and embedded processors in the base plate and implement secondary acceleration using configurable metal layers.

[00603] Additionally, as discussed above, a chip implemented as an ISSN, FPGA, or structured ISSN may include or otherwise be associated with one or more memory architectures. For example, a memory architecture may consist of M memory modules that interact with a chip, such as an ISSN. ISS can be implemented using a variety of different technologies, including FPGA (Field Programmable Gate Arrays) or structured ISS, standard cells, or completely custom logic. The ISSN contains a memory subsystem (MSS) and functional processors (FP). The MSS contains M memory controllers (MCs) for memory modules, N system memory interfaces (SMIs) for FPs, and N×M switches that allow any SMI to access any MC. In case of conflicts, conflict resolution is ensured.

[00604] Each memory module is constructed of DRAM chips that can be accessed by an A _MM bit word and support data transfer of a width of D _MM bits. The specified memory has address locations. The key characteristic of DRAM is that it performs read/write bursts of W words using the passed address as the base address, B, and retrieving or storing locations B+1, B+2, ... B+W-1. Typically W is 8.

[00605] In the MSS ISSN, each memory controller provides the required control signal and performs multiplexing/demultiplexing as necessary between the system word width, D _SYS , and the memory word width, D _MM , and also fulfills the requirements for read/write packets. It can have additional buffering so that multiple memory requests can be queued and processed in a pipelined manner to maximize throughput. This compensates for the many clock cycles of delay between providing the address and completing the memory operation (writing or reading).

[00606] The MC can operate at the speed of the attached DRAM in the memory module. Let's assume that its clock speed is C _MM . It is typically several times faster than the core speed at which most of the logic in ISSN runs, which is C _SYS . Therefore, the multiplexing/demultiplexing logic is located close to its associated interface pins to minimize the distances the signals travel. Demultiplexing is the first operation performed on incoming data, and multiplexing is the last operation performed on outgoing data. The remainder of the MSS operates on data with a bit width of D _SYS that is larger than D _MM , allowing the use of the slower C _SYS clock rate.

[00607] Each system memory interface in the MSS represents an A _SYS -bit address bus and a D _SYS -bit data bus for any attached FP. SMI is designed so that the attached FP appears to have random access to a single, high-capacity, fast memory. The FP is not aware of the existence of individual memory modules. The A _SYS value is large enough to allow access to the memory location in any attached memory module. Mapping from the system address space to the memory module address space is described below.

[00608] The N system memory interfaces are cross-connected to the M memory modules via an N×M switch. The specified switch provides min(M,N) simultaneous connections between SMI and MC, provides conflict resolution and facilitates the translation of the system address space into the memory module address space.

[00609] The organization of the FP is largely flexible. One or more FPs may store the same system memory interface. To maximize performance, FPs that are not running at the same time should share an SMI. And those that work simultaneously must be connected to different SMIs. An FP that operates on data structures larger than D _SYS can use multiple SMIs to access the entire data structure in a single memory operation. Consequently, the specified data architecture supports a wide range of computing architectures. Each FP can be identical and, therefore, an array of them can be implemented in a two-dimensional structure. This is illustrated when FPU(i,j) is the jth component attached to SMI i, 0≤i<N, 0≤j<k _i . In this case, all k _i have the same size, and k _i can be small and equal to 1. This supports SIMD (single instruction stream, many data streams) and MIMD (multiple instruction stream, multiple data stream) architectures, depending on whether whether FP commands are from the same or separate command memories.

[00610] In one particular aspect, the present invention is directed to a system, such as a system for executing a sequence analysis pipeline on genetic sequence data. In various examples, the system may comprise an electronic data source, such as a data source that provides digital signals, such as digital signals representing a plurality of genomic data reads, each of the plurality of genomic data reads comprising a nucleotide sequence. The system may include one or more storage devices, such as a memory, storing one or more control genetic sequences and/or an index of the one or more control genetic sequences; and/or the system may include a chip such as an ISSN, FPGA, or CCIS.

[00611] More specifically, in various specific embodiments, the system may comprise a special-purpose structured integrated circuit (SPIC), such as one in which the chip is formed by a set of hard-wired mask-programmable digital logic circuits that can be interconnected through a plurality of physical electrical interconnects. . In various examples, one or more of the plurality of physical electrical interconnects comprises an input to a structured ISSN that is coupled to an electronic data source, for example, to receive a plurality of genomic readom data. In such an example, one or more of the plurality of physical electrical interconnections may comprise a memory interface for structured ISS access to memory. Accordingly, the hard-wired digital logic circuits may be implemented as a set of processing modules, such that each processing module may be formed by a subset of the hard-wired digital logic circuits to perform one or more steps in a sequence analysis pipeline on a plurality of genomic data reads. For example, one or more, such as each, of subsets of hard-wired digital logic circuits may be configured to perform one or more steps in a sequence analysis pipeline. For example, a set of processing modules may be configured to contain one or more of a mapping module, an alignment module, and/or a sorting module.

[00612] For example, a set of processing modules may include a mapping module in a mounting configuration configured to access, in accordance with at least a portion of a nucleotide sequence in a read of a plurality of reads, an index of one or more reference genetic sequences from memory via an interface memory for mapping the read into one or more segments of one or more reference genetic sequences based on the index. For example, in some embodiments, an index of one or more reference genetic sequences may comprise a hash table and/or a mapping module may apply a hash function to at least a portion of the nucleotide sequence to access the hash table of the index.

[00613] The processing engines may also, or alternatively, comprise an alignment module in a mounting configuration configured to access one or more reference genetic sequences from memory, such as through a memory interface, to align the read to one or more positions in one or more segments of one or more control genetic sequences, such as those obtained from the mapping module. The processing engines may also or alternatively comprise a sort module in a mounting configuration configured to access one or more aligned reads from memory, such as through a memory interface for sorting the read into one or more positions, such as chromosome positions, in one or more control genetic sequences, such as those obtained from the alignment module.

[00614] In various examples, the structured ISS may comprise a base plate that contains at least some of the hard-wired digital logic circuits, and in some examples may include one or more configurable metal layers formed on the base plate, such as where each of one or The more configurable metal layers may have at least some of a plurality of physical electrical interconnections that provide interconnection of at least some of the hardwired digital logic circuits to form a set of processing modules. In some embodiments, one or more of the plurality of physical electrical interconnections may comprise output from the structured ISSN, for example, to carry resultant data from a mapping module, an alignment module, and/or a sorting module.

[00615] In various examples, the structured ISS may include a master control device for establishing a wiring configuration for each subset of hard-wired digital logic circuits to perform one or more steps in the sequence analysis pipeline. In various embodiments, the mounting configuration is made during integrated circuit manufacturing and is non-permanent. In some embodiments, the structured ISSN and/or storage device is located in an expansion card, such as a peripheral component interface (PCI) card. As discussed above, in various embodiments, the system may comprise a sequencer, such as one that contains an electronic data source that provides digital signals representing multiple reads of genomic data. And in such an example, the expansion card can be physically integrated into the sequencer.

[00616] Additionally, in various embodiments, a special purpose structured integrated circuit (SPI) may be provided, such as for analyzing genetic sequence data, such as when the genetic sequence data is stored in a storage device, such as storing one or more control genetic sequences associated with genomic data, and/or an index of one or more reference genetic sequences. In such an example, the structured ISS may include a base plate that also contains a set of digital logic circuits, and may also include one or more configurable metal layers formed on the base plate, for example where each of the one or more configurable metal layers may have a set of wiring connections, which provide interconnection of a subset of digital logic circuits to form a set of processing modules. In such an example, the set of processing modules may comprise a mapping module, an alignment module, and/or a sorting module. In various examples, a portion of a set of digital logic circuits in the base plate is hardwired as a base sensing module.

[00617] For example, a set of processing modules may include a mapping module for accessing, in accordance with at least a portion of a nucleotide sequence in a read of a plurality of reads, an index of one or more reference genetic sequences stored in memory for mapping the read into one or more segments of one or more control genetic sequences, for example based on an index. Additionally or alternatively, the set of processing modules may include an alignment module for accessing one or more reference genetic sequences from memory, for example through a memory interface, to align the read to one or more positions in one or more segments of the one or more reference genetic sequences from the module mapping. Additionally or alternatively, the set of processing modules may include a sorting module for sorting each aligned read according to one or more positions in one or more reference genetic sequences.

[00618] In one embodiment, a system is provided for implementing a sequence analysis pipeline over genetic sequence data, the system comprising an electronic data source that provides digital signals representing a plurality of genomic data reads, each of the plurality of genomic data reads comprising a nucleotide sequence. The system may include one or more storage devices, for example for storing one or more control genetic sequences; and/or the system may comprise an integrated circuit having a base plate, for example a base plate formed by a photolithographic mask that defines a set of digital logic circuits. In such an example, the base plate may be configured to have one or more functions, as described above herein, integrated therein. For example, the base plate may have one or more configurable metal layers, such as where each of the one or more configurable metal layers has one or more interconnects that connect a subset of a set of digital logic circuits in a mounting configuration to perform the above functions.

[00619] In accordance with various aspects, a method is provided for implementing a special purpose structured integrated circuit (SIC) for analyzing genetic sequence data. In some embodiments, the method includes one or more of providing a plurality of photolithographic masks, such as masks that define a set of digital logic circuits of the base plate; forming a set of digital logic, for example using a plurality of photolithographic masks to form a base plate; providing two or more different sets of mask design-specific configurable metal layers, such as masks that define corresponding two or more digital logics for implementing a set of processing modules; forming two or more configurable metal layers, for example using two or more mask design-specific configurable metal layers, for example wherein each of the two or more configurable metal layers has a set of mounting connections that can be made in accordance with the design of the configurable metal layers masks, for example, to interconnect a subset of digital logic circuits to form sets of processing modules; and/or providing two or more configurable metal layers on the base plate to form a set of processing modules.

[00620] According to one or more aspects or features of the subject matter described herein, may be implemented in a digital electronic circuit, an integrated circuit, specially designed special purpose integrated circuits (SIICs), field programmable gate arrays (FPGAs), or a structured ISS, computer hardware, firmware, software and/or combinations thereof.

[00621] These various aspects or features may include implementation in one or more computer programs that can be executed and/or interpreted on a programmable system, including at least one programmable processor, which may be specialized or general purpose, associated with the ability to receive data and commands from the storage system, at least one input device and at least one output device, and transmitting data and commands thereto. The programmable system or computing system may include clients or servers. The client and server are usually remote from each other and typically communicate via a communications network. The client-server relationship occurs due to computer programs running on the respective computers and having a client-server relationship with each other.

[00622] These computer programs, which may also be referred to as programs, software, software applications, applications, components, or code, contain machine instructions for a programmable processor and may be implemented in a procedural and/or high-level object-oriented programming language and/or in assembly/machine language. As used herein, the term “computer readable medium” refers to any computer program product, instrument and/or device, such as, for example, magnetic disks, optical disks, storage devices, and programmable logic devices (PLDs) used to provide machine instructions. and/or data into a programmable processor, including a computer-readable medium, that receives the machine instructions as a computer-readable signal. The term "machine-readable signal" refers to any signal used to provide machine instructions and/or data to a programmable processor. A computer-readable medium may store such machine instructions in a non-transitory manner, such as a non-transitory solid-state storage device, a hard disk drive, or any equivalent storage medium. A computer-readable medium may alternatively or additionally store such machine instructions in a transient manner, such as a processor cache or other random access storage device associated with one or more physical processor cores.

[00623] To provide user interaction, one or more aspects or features of the subject matter described herein may be implemented on a computer having a display device, such as, for example, a cathode ray tube (CRT), liquid crystal display (LCD), or a light-emitting diode (LED) monitor for displaying information to the user, and a keyboard or pointing device, such as a mouse or trackball, with which the user can input into the computer. Other types of devices can also be used to provide user interaction. For example, the user feedback may be any form of sensory feedback, such as, for example, visual feedback, audio feedback, or haptic feedback; and user input may be received in any form, including, but not limited to, audio, speech, or tactile input. Other possible input devices include, but are not limited to, touch screens or other touch devices such as single or multi-point resistive or capacitive touch panels, voice recognition hardware or software, optical scanners, optical pointers, digital image capture devices with associated software provision for interpretation, etc.

[00624] The subject matter described herein may be implemented in systems, devices, methods and/or products depending on the desired configuration. The embodiments given in the foregoing description do not represent all embodiments corresponding to the subject matter described herein. On the contrary, these are just some examples corresponding to aspects related to the described subject matter of the invention. Although several options are described in detail above, other modifications and additions are possible. In particular, the features and/or options described herein may be supplemented by other features and/or options. For example, the embodiments described above may relate to various combinations and subcombinations of the described features and/or to combinations and subcombinations of several additional features described above. In addition, the logical flows shown in the accompanying figures and/or described herein do not necessarily require adherence to the particular order shown or sequential order to achieve the desired results. Other embodiments are possible within the scope of the attached claims.

Claims (54)

1. A platform for analyzing genomic data, wherein said platform for analyzing genomic data comprises: a graphical user interface that presents a plurality of user selectable options, where one or more of said plurality of user selectable options corresponds to a particular genomic data processing pipeline; And application programming interface (API) of the platform, which (i) receives data representing a selection of one or more of a plurality of user selectable options corresponding to a particular genomic data processing pipeline, and (ii) configures one or more programmable logic device circuits to execute a set of one or more genomic data processing pipelines based on the received data representing the selection, where the configuration includes defining by an application programming interface (API) inputs for each of the one or more processing pipelines genomic data executed by said configured programmable logic device circuits based on the received data representing the selection. 2. The genomic data analysis platform of claim 1, wherein one or more of the plurality of user selectable options correspond to a specific genomic data analysis application that is stored in the one or more application repositories; And wherein the API further (iii) receives second data representing a selection of one or more user selectable options corresponding to a particular genomic data analysis application, and (iv) configures one or more inputs for corresponding genomic data analysis applications stored in one or more repositories applications based on the obtained second data. 3. The genomic data analysis platform of claim 1, wherein said plurality of user-selectable options includes one or more of a user-selectable genome pipeline option, a user-selectable epigenome pipeline option, a user-selectable metagenome pipeline option, a user-selectable option corresponding to the co-genotyping processing pipeline, or a user-selectable option corresponding to the genomic analysis software (GATK) processing pipeline. 4. The genomic data analysis platform of claim 2, wherein said plurality of user-selectable options includes one or more of a user-selectable option corresponding to a non-invasive prenatal testing application, a user-selectable option corresponding to a neonatal intensive care unit application, a user-selectable option corresponding to an application cancer assay, a user-selectable option corresponding to a laboratory-based research (LDT) application, or a user-selectable option corresponding to an agricultural and biological analysis application. 5. A method for analyzing genomic data, including: obtaining, through an application program interface (API) implemented by one or more computers, first data representing a selection of one or more user selectable options input via a graphical user interface, wherein one or more of the plurality of user selectable options identifies a particular genomic processing pipeline data; configuring, using an API implemented by one or more computers, programmable logic device circuits to execute a genomic data processing pipeline based on said first data, wherein configuring the genomic data processing pipeline includes using the API to define each of the one or more genomic data processing pipelines identified first data and executable by said configured programmable logic device circuits; obtaining, by one or more computers, second data representing a set of genomic data or a set of data derived from genomic data; using, by said one or more computers, said genomic data processing pipeline configured in said programmable logic device circuits, based on the first data, to process the obtained second data; obtaining, by said one or more computers, resultant data generated by said genomic data processing pipeline configured in programmable logic device circuits based on the genomic data processing pipeline's processing of the received second data; And providing, by said one or more computers, output data based on said result data. 6. The method of claim 5, wherein said genomic data set includes one or more genomic sequences generated by a nucleic acid sequencer. 7. The method of claim 5, wherein said set of data derived from genomic data includes a set of one or more options. 8. Method according to claim 5, where one or more of the plurality of user selectable options correspond to a particular genomic data analysis application stored in one or more application repositories, wherein said first data further includes data representing a selection of one or more user selectable options, each of which identifies one or more genomic data analysis applications, and the method additionally includes: configuring, via an API, inputs for one or more corresponding genomic data analysis applications stored in one or more application repositories based on the first data obtained. 9. The method of claim 5, wherein said plurality of user selectable options includes one or more of a user selectable option corresponding to the genome processing pipeline, a user selectable option corresponding to the epigenome processing pipeline, a user selectable option corresponding to the metagenome processing pipeline, a user selectable option, corresponding to the co-genotyping processing pipeline, or a user-selectable option corresponding to the genomic analysis software (GATK) processing pipeline. 10. The method of claim 9, wherein said plurality of user selectable options includes one or more of a user selectable option corresponding to a non-invasive prenatal testing application, a user selectable option corresponding to a neonatal intensive care unit application, a user selectable option corresponding to a cancer analysis application, selectable a user-selectable option corresponding to a laboratory-based research (LDT) application or a user-selectable option corresponding to an agricultural and biological analysis application. 11. A non-transitory computer-readable storage medium for genomic data analysis on which software is stored containing instructions that can be executed by one or more computers which, when so executed, cause said one or more computers to perform operations including: receiving, through an application program interface (API), first data representing a selection of one or more user selectable options input via a graphical user interface, where one or more of the plurality of user selectable options identifies a particular genomic data processing pipeline; configuring, using said API, programmable logic device circuits to execute a genomic data processing pipeline based on said first data, where configuring the genomic data processing pipeline includes using the API to define inputs to each of the one or more genomic data processing pipelines identified by the first data, and executed by programmable logic device circuits; obtaining second data representing a set of genomic data or a set of data derived from the genomic data; using said genomic data processing pipeline configured in programmable logic device circuits based on the first data to process the obtained second data; obtaining result data generated by said genomic data processing pipeline configured in the programmable logic device circuits based on the genomic data processing pipeline's processing of the received second data; And providing output that is based on specified output data. 12. The computer-readable medium of claim 11, wherein said genomic data set includes one or more genomic sequences generated by a nucleic acid sequencer. 13. The computer-readable medium of claim 11, wherein said set of data derived from genomic data includes a set of one or more variants. 14. Machine-readable medium according to clause 11, where one or more of the plurality of user-selectable options correspond to a particular genomic data analysis application stored in one or more application repositories, wherein said first data further includes data representing a selection of one or more user selectable options, each of which identifies one or more genomic data analysis applications, and Moreover, these operations additionally include: configuring, via an API, inputs for one or more corresponding genomic data analysis applications stored in one or more application repositories based on the first data obtained. 15. The computer-readable medium of claim 11, wherein said plurality of user-selectable options includes one or more of a user-selectable option corresponding to a genome processing pipeline, a user-selectable option corresponding to an epigenome processing pipeline, a user-selectable option corresponding to a metagenome processing pipeline, a user-selectable option , corresponding to the co-genotyping processing pipeline, or a user-selectable option corresponding to the genomic analysis software (GATK) processing pipeline. 16. The computer-readable medium of claim 14, wherein said plurality of user selectable options includes one or more of a user selectable option corresponding to a non-invasive prenatal testing application, a user selectable option corresponding to a neonatal intensive care unit application, a user selectable option corresponding to a cancer analysis application, a user-selectable option corresponding to a laboratory-based research (LDT) application, or a user-selectable option corresponding to an agricultural and biological analysis application. 17. System for analyzing genomic data, containing: one or more computers and one or more storage devices storing instructions that, when executed by said one or more computers, can cause said one or more computers to perform operations including: obtaining, through an application programming interface (API) hosted on said one or more computers, first data representing a selection of one or more user selectable options input via a graphical user interface, wherein one or more of the plurality of user selectable options identifies a particular pipeline processing of genomic data; configuring, using said API hosted on said one or more computers, programmable logic device circuits to execute a genomic data processing pipeline based on said first data, wherein configuring the genomic data processing pipeline includes using the API to define inputs to each of the one or more processing pipelines genomic data identified by the first data and executed by programmable logic device circuits; obtaining, by one or more computers, second data representing a set of genomic data or a set of data derived from genomic data; using, by said one or more computers, said genomic data processing pipeline configured in programmable logic device circuits, based on the first data, to process the obtained second data; obtaining result data generated by said genomic data processing pipeline configured in the programmable logic device circuits based on the genomic data processing pipeline's processing of the received second data; And providing, by said one or more computers, output data based on said result data. 18. The system of claim 17, wherein said genomic data set includes one or more genomic sequences generated by a nucleic acid sequencer. 19. The system of claim 17, wherein said set of data derived from genomic data includes a set of one or more variants. 20. The system of claim 17, wherein one or more of the plurality of user selectable options corresponds to a particular genomic data analysis application stored in one or more data stores, wherein said first data further includes data representing a selection of one or more user selectable options, each of which identifies one or more genomic data analysis applications, and wherein said operations further include: configuring, via an API, inputs for one or more corresponding genomic data analysis applications stored in the one or more application stores, based on the first data obtained. 21. The system of claim 17, wherein said plurality of user selectable options includes one or more of a user selectable option corresponding to the genome processing pipeline, a user selectable option corresponding to the epigenome processing pipeline, a user selectable option corresponding to the metagenome processing pipeline, a user selectable option, corresponding to the co-genotyping processing pipeline, a user-selectable option corresponding to the genomic analysis software (GATK) processing pipeline. 22. The system of claim 20, wherein said plurality of user selectable options includes one or more of a user selectable option corresponding to a non-invasive prenatal testing application, a user selectable option corresponding to a neonatal intensive care unit application, a user selectable option corresponding to a cancer analysis application, selectable a user-selectable option corresponding to a laboratory-based research (LDT) application or a user-selectable option corresponding to an agricultural and biological analysis application.