VARIANTMEDIUM

Toolbox for mutation calling using deep learning

Dependencies

• nextflow >= 24.10.3
• conda >= 4.4 (miniconda >=23.11.0 recommended)
• CUDA 11.4 (optional for GPU support)

Installation

git clone https://github.com/TRON-Bioinformatics/variantmedium.git
cd variantmedium

Build conda environments

bash build.sh config.conf

Usage

First, you will need to state the paths to your tumor-normal BAM pairs in a tab separated file in the following format without header. All fields are in string format, so you are free to choose the name and replicate_pair_identifier as long as they make a unique tuple.

sample name	replicate pair identifier	tumor bam path	normal bam path
sample_1	rep_1	tumor_1.bam	normal_1.bam
sample_2	rep_2	tumor_2.bam	normal_2.bam

Define the following variables in config.conf:

• CODE_FOLDER Directory of the VariantMedium source code
• ENV_FOLDER Directory for conda envirnments
• REF_FOLDER Directory for reference data
• PAIRS File path to tab-separated table with samples associated to tumor/normal BAM files
• OUT_FOLDER Output folder
• KNOWN_INDELS1 Common indel variant file for BAM preprocessing pipeline
• DBSNP dbDNP VCF file for BAM preprocessing pipeline
• REF Reference genome
• EXOME_BED Target region defintion as BED file (e.g. exome) - Leave empty ("") if calling in WGS

If you need to apply BAM-Preprocessing and need the resource files, you can download the full reference data for the human genome hg38, run 'sh download_ref.sh` and use configurations as defined in 'config_hg38.conf' for the reference genome GrcH38. This script also provides download commands for reference genome and S07604624 SureSelect Human All Exon V6+UTR from UCSC if you need them.

Please make sure the index for the BAM file exists with the ".bai" extension under the same directory, e.g. for bams/tumor.bam you have bams/tumor.bai. Also please make sure that the tumor and normal bam files do not have the same file name, even if they are under different directories. ( Linking with a new file name is ok)

Finally, run the VariantMedium pipeline by

bash run.sh config.conf

Output

You will then find the calls in your OUT_FOLDER as tsv and VCF files. (<sample_name> .somatic_snv.VariantMedium.vcf/<sample_name>.somatic_snv.VariantMedium.tsv) The variants are sorted by the neural network score.

Errors running the pipeline

We listed the solutions to common errors we encountered when running this pipeline under troubleshooting.md document in this repository

Training data

We share the cell-line sequencing data and orthogonal deep sequencing confirmation of variants under controlled access. The data is avaliable under European Genome-Phenome Archive (EGA) with accession number EGAS00001007633

Citation

A manuscript describing the method will be available soon.

License

• The source code is distributed under a MIT license
• The parts of the source code that use torchvision are distributed under BSD 3-Clause License
• The machine learning models in the folder models are distributed under CC BY-NC-ND 4.0