Avenger is a programming language to aid in both AI and Humans to write more scientifically accurate code. Avenger is a Lua dialect that uses cpython, making the entire world of python3 libraries available to Avenger. Avenger simplifies code by having a large common-core of libraries that you should already expect to be included by your program. Making Avenger code easier to read, and prevents incompatible code due to irregular includes that can occur in python.

Avenger is ideal for rapid development of bioinformatics pipelines, for hackathons where time is important. Or for homework assignment where you need reproducible results from code that is intelligible and easy to maintain and reuse.

This project simplifies the process of building plasmids and packaging them up into a single annoated genbank file that can be easily used on benchling or another workbench. Although this task is commonly perfomred by hand in a browser or desktop application, performing senstive tasks like this with python allows us to be eaxct with our changes and to verify the changes.

The project aims to be a swiss army kinfe of common pipelines used in gene editing. The current pipeline consists of three main steps: (1) downloading gene data from NCBI, (2) discovering mutations within the genome using GATK4, and (3) predicting off-targets for a set of gRNA candidates using DeepCRISPR's DCModelOntar model.

• Python 3.7+
• Conda
• GATK4
• tenserflow 1.3.2 (todo: tf2 update)

Clone the repository

git clone https://github.com/SweetWaterAI/PlasmidCompiler

Install required packages using the command below

pip install -r requirements.txt

Add GATK4 to the path

export PATH=$PATH:/path/to/gatk4

This function downloads gene data for a given gene name from NCBI and saves all fields as annotations in a genbank file. If the gene data has already been downloaded, the function exits without re-downloading the data.

$ python main.py get <gene_name> [--email <email>]

• gene_name (required): Name of the gene to download.
• email (optional): Email address to be used for NCBI queries.

This function uses GATK4 to find all mutations of a given gene within a genome. The function requires a reference genome file, the name of the gene of interest, and a path to the output VCF file.

$ python main.py discover <reference_file> <gene_name> <output_file>

• reference_file (required): Path to the reference genome file.
• gene_name (required): Name of the gene of interest.
• output_file (required): Path to the output VCF file.

This function predicts off-targets for a set of gRNA candidates and target sequences using DeepCRISPR's DCModelOntar model. The function requires the path to the gRNA candidates file, the path to the target sequences file, the path to the full genome sequence file, and the path to the DCModelOntar model file.

$ python main.py predict <gsRNA_candidates> <targets> <genome> [--output <output>] --model_path <model_path> [--pam <pam>]

• gsRNA_candidates (required): Path to a file containing gRNA sequences.
• targets (required): Path to a file containing target sequences.
• genome (required): Path to the full genome sequence.
• output (optional): Path to the output file. Default is "ontar_results.tsv". model_path (required): Path to the DCModelOntar model file.
• pam (optional): PAM sequence used by the gRNA. Default is "NGG".

Commands can be saved within a .sh file and re-run to re-build a specific plasmid when needed. A team should be able to build any number of therapies or tests out of the same directory. The code will take in fasta or genebank files.

This method produces GenBank file of a plasmid backbone by replacing specified target genes with new sequences. It takes in five arguments:

$ python main.py build --backbone-file path/to/backbone.gb --target-dir path/to/targets --replacement-dir path/to/replacements --crispr-file path/to/crispr.gb --output-file path/to/output.gb

• backbone_file (required): the path to the GenBank file of the plasmid backbone to build from
• target_dir (required): the path to the directory containing the list of target genes to replace
• replacement_dir (required): the path to the directory containing the list of replacement genes to use
• crispr_file (required): the path to the GenBank file containing the desired multiplexed-CRISPR variant
• output_file (required): the path to the resulting plasmid

The function first parses the CRISPR GenBank file to get the target sequence and PAM. It then parses the GenBank file of the plasmid backbone. For each target gene and its replacement, the function finds the location of the target gene in the plasmid backbone. It then replaces the target gene with the replacement sequence. If it cannot find the target gene in the plasmid backbone, the function raises a ValueError.

Finally, the function writes the modified plasmid GenBank file to disk.