One stop MIP design and analysis
Use MIPgen to design custom mip panels for target enrichment of moderate to high complexity DNA targets ranging from 120 to 250bp in size
To compile MIPgen, you will need a C++ compiler, such as GCC (https://gcc.gnu.org/). That's it! For running design and analysis, see below for dependencies
To build MIPgen from source, simply enter the MIPGEN directory and type 'make'
The 'mipgen' executable can be used to perform designs
MIPgen accepts many options intended to maximize the flexibility of your designs
The only required parameters are the minimum and maximum sizes for the captured sequence, a prefix for output files and a BED file consisting of the coordinates of the desired targets, and the bwa- and samtools-indexed reference genome from which to pull the sequences
Other options provide control over tiling, scoring and oligo design behavior
MIPgen offers three scoring methods based on two methods of scoring: SVR and logistic regression. Both methods offer similar performance, with slight improvements seen with the SVR scoring. The 'mixed' scoring option will perform designs primarily with the faster logistic regression score and finish with the more accurate SVR scoring scheme.
Run ./mipgen -doc to see the full documentation or read the text below
-It is not recommended to include more than 8bp total of degenerate smMIP tags. Longer tags reduce specificity and complicate MIP rebalancing, leading to less complete coverage of targeted sites.
-Priority scores (used by default) modulate tiling behavior and density of tiling of low scoring regions. To reduce density, lower priority scores. In particular, for interrogation of very short loci (<10bp), priority scores should be set to 0 to prevent splitting the region into two MIP targets.
-Regions that do not map uniquely to the provided reference are not tiled by default. These and other bases that cannot be tiled will be reported in the <project_name>.coverage_failed.bed file.
-SVR scores below 1.4 and logistic scores below 0.7 rarely perform adequately. It is recommended to remove these probes from designed assays.
-The failure flags in the design files are a series of three bits. They represent, in order, non-uniqueness of the MIP target with respect to the provided genome sequence, inability to design SNP MIPs to any/all of the variants provided in the SNP file, and tandem repeats being detected in the MIP arms using Tandem Repeats Finder. SNP failures will be included in final probe sets. The others will not, assuming non-uniqueness is selected and a TRF executable is provided
-MIPgen does not remove candidate MIPs with overly high targeting arm copy numbers if there are no suitable alternatives to cover the desired regions. Be advised that copy numbers above 200 are likely to poisson your assay and should be removed unless you are specifically interrogating all such low complexity sites.
-the subset of Boost required for compilation is included in the directory
-the makefile should compile the source without any external dependencies, but the executable that is generated makes system calls to SAMtools, BWA, Tabix, and, optionally, Tandem Repeats Finder
-the tools subdirectory has scripts for parsing fastqs prior to mapping and collapsing tag-defined read groups (TDRGs) and trimming targeting arms after mapping (Cython required)
-summary information is also printed as part of the collapsing process and requires NumPy and SciPy
-if there is at least 10bp of overlap between forward and reverse reads, it is recommended to merge read pairs into fr-reads using PEAR ( https://bioinformatics.oxfordjournals.org/content/early/2013/11/10/bioinformatics.btt593.long ) prior to processing and to utilize single end options
-see the HemoMIPs 2020 analysis pipeline for analyzing MIP sequencing data with html reports and full genotyping results: https://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1007956
Below is an example for getting MIPgen up and running and designing probes to EGFR, TERT and BRAF. The output generated from these commands is included in the repository (mipgen_example.tgz). A copy of the human reference genome (GRCh37, also known as hg19 in UCSC) is available at the 1000Genomes FTP site:
ftp:https://ftp.1000genomes.ebi.ac.uk/vol1/ftp/technical/reference/human_g1k_v37.fasta.gz
The file used in the example to pull out gene exons for the hg19 human reference (refGene.txt) can be downloaded here:
https://hgdownload.cse.ucsc.edu/goldenPath/hg19/database/
First extract the tar file to make a MIPgen directory
tar -xf mipgen.tgz
Enter the directory to compile the source code
cd MIPGEN/
Use make to compile -- make sure you have a C++ compiler
make
Make a directory for practicing designs
mkdir ../mipgen_practice
cd ../mipgen_practice
Create a new file and type a list of gene symbols you would like to tile with MIPs
vim practice_genes.txt
You can use one of the scripts in the tools subdirectory and a refGene text file listing gene exons to pull out the genomic coordinates of your genes of interest
sh ../MIPGEN/tools/extract_coding_gene_exons.sh practice_genes.txt ../refGene.txt > practice_genes.bed
Check to make sure the coordinates are what you expect and that no errors occurred
less practice_genes.bed
Now you can perform designs with MIPgen! Here is a design with very basic options. Set the path to the genome reference and change the name of the FASTA file if yours is different. Make sure you have the dependencies installed or accessible through a given path (BWA, tabix, samtools)!
../MIPGEN/mipgen -regions_to_scan practice_genes.bed -project_name practice_design -min_capture_size 162 -max_capture_size 162 -bwa_genome_index /<path to genome reference fasta file>/human_g1k_v37.fa
The final selection of MIPs is located in the picked MIPs file review the scores to make sure the MIPs stand a good chance of success (logistic scores below 0.6 are unlikely to provide usable data)
less practice_design.picked_mips.txt
By default all tested MIPs are output; this is a lot of output! (Turn it off with the silent_mode option)
rm practice_design.all_mips.txt
sai, fq (and sam) files are not deleted automatically and can also take up space
rm -f *.sai *.fq
Generate a UCSC track with another tools script to visualize online
python ../MIPGEN/tools/generate_ucsc_track.py practice_design.picked_mips.txt practice_ucsc_track
Look at the other files in this directory to see designs for TERT, BRAF and EGFR (we have not tested these designs experimentally so we cannot precisely assess predicted performance for these probes)
usage: mipgen (<-parameter_name> <parameter_value>)*
Created by Evan Boyle ([email protected])
Required parameters:
-project_name prefix for output
ex: /my/output/directory/batch_1
-bwa_genome_index samtools- and bwa-indexed genome reference file path
-regions_to_scan BED file of positions GUARANTEED to be captured
regions will be merged as needed
-min_capture_size integer value for length of targeting arms plus insert region to be captured
tested down to 120
-max_capture_size integer value for length of targeting arms plus insert region to be captured
tested up to 250
Oligo control:
-arm_lengths manual setting of individual possible ligation and extension arm lengths
format: <extension arm length 1>:<ligation arm length 1>,<extension arm length 2>:<ligation arm length 2>,...
ex: 16:24,16:25,16:26
-arm_length_sums setting of arm lengths to include all pairs of arm lengths adding to a certain length
default is 40,41,42,43,44,45
format: <sum of extension and ligation arm lengths 1>,<sum of extension and ligation arm lengths 2>,...
ex: 40,45
-ext_min_length minimum length of extension arm
default is 16
-lig_min_length minimum length of ligation arm
default is 18
-tag_sizes specifies degenerate tag (Ns) to place in extension arm and ligation arm
more than 8 bases severely reduces on-target rate of capture
default is 5,0
format: <extension arm tag size>,<ligation arm tag size>
ex: 4,4
-masked_arm_threshold fraction of targeting arms having masked bases (floating point number from 0 to 1)
to tolerate before optimizing with respect to masking (requires TRF executable)
default is 0.5
-target_arm_copy threshold over which minimizing copy number to the reference takes priority
default is 20
-max_arm_copy_product maximum permissible product of targeting arm copy number to the reference
default is 75
Tool dependencies:
-tabix tabix build
default is "tabix"
-bwa bwa build
default is "bwa"
-trf tandem repeats finder executable
default is "off"
providing a TRF executable enables filtering of simple repeats in probe arms
Input options:
-genome_dir genome reference directory helpful for large jobs
ex: /my/genome/split/by/chromosomes/
-snp_file path to vcf file of snps to avoid during design
-common_snps providing "off" will disable loading of common SNPs to avoid from NCBI using Tabix
DEPRECATED:ONLY SNP FILE OPTION NOW SUPPORTED
-file_of_parameters file containing any of the above parameters in the following format:
-first_parameter_name first_parameter_value
-second_parameter_name second_parameter_value
<etc.>
lines that do not start with '-' are ignored
Tiling control:
-feature_flank integer value for extending BED coordinates
default is 0
-capture_increment integer step size for examining capture sizes
default is 5
starts at maximum and descends until minimum is reached or
acceptable score found, whichever is first
-logistic_heuristic providing "off" will remove assumptions to reduce search space
-max_mip_overlap integer specifying the maximum number of nucleotides of overlap
to test before selecting a MIP
default is 30
-starting_mip_overlap integer specifying the starting number of nucleotides of overlap
to use when linearly tiling with MIPs
default is 0
-check_copy_number providing "off" enables targeting of non-unique sites in the reference
-seal_both_strands providing "on" disallows targeting arms occupying the same base
positions on opposite strands
-half_seal_both_strands providing "on" disallows a second targeting arm from occupying more than half
of the positions of a targeting arm placed on the other strand
-double_tile_strand_unaware providing "on" will ensure that all positions in targeted regions will be tiled twice
-double_tile_strands_separately providing "on" will ensure that all positions in targeted regions will be tiled twice,
at least once on each strand
Scoring parameters:
-score_method "logistic" will use the logistic model for scoring (default)
"svr" will switch behavior to score with the svr model
"mixed" will first filter with logistic scoring and finalize with the svr model
-logistic_optimal_score value of score metric at which to stop optimizing MIP score and
accept current MIP, lower scores lead to fewer outliers
default is 0.98
-svr_optimal_score analogous to logistic_optimal_score
default is 2.2
-logistic_priority_score value of score metric below which MIPs are placed nonlinearly to
optimize score and SNPs are tolerated
default is 0.9,lower values enable more efficient linear tiling
-svr_priotity_score analogous to logistic_priority_score
default is 1.5 for SVR
Miscellaneous:
-silent_mode providing "on" will reduce volume of text output
-download_tabix_index providing "on" will force redownload of common snp tbi file
DEPRECATED
-bwa_threads make use of BWA's multithreading option (-t _)
default is 1
Copyright 2014 University of Washington.
Permission is hereby granted, to You to use, copy, modify, merge the MIPgen software and associated documentation files (the "Software") for the use in Your non-commercial academic and research activities. This includes electronic distribution for the use of client software, but only within Your organization. You may not otherwise distribute the Software, including any modified versions. You may not sublicense any rights under this license and You may not sell copies of the software. The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
Contact [email protected] for any uses not covered under this license.
© University of Washington 2014