svigneau / variant_caller Goto Github PK

Variant calling pipeline that takes paired end short read data (fastq files) and a reference genome (fasta file) as inputs and output variant calls (vcf file). The pipeline uses snakemake, bwa, samtools, and freebayes.

Before running the pipeline:

• add paired end short read fastq files to the fastq/ directory
• add reference genome fastq file to the ref/ directory
• edit config/config.yaml file with the base name and extension of the fastq files and the name of the fasta file

The pipeline can be run with one of the following commands:

• snakemake --cores [number of cores] uses local dependencies
• snakemake --cores [number of cores] --use-conda uses snakemake integrated package management to fetch dependencies from conda repository and create self-contained environments
• snakemake --cores [number of cores] --use-conda --use-singularity uses snakemake integrated package management within a Docker container created from DockerHub (docker://svigneau/variant_caller) and carrying the required dependencies

Notes:

Unit tests generated with snakemake --generate-unit-tests were modified to do the following:

• handle config and log files as described in: snakemake/snakemake#843 (comment)
• remove date line from vcf files

The unit tests use SARS-CoV-2 reference genome (MN908947.3) and sequencing data from a Covid patient (downsampled from SRR15660643). In the resulting vcf files, most polymorphisms are homozygous as expected with a haploid viral genome. However, a small proportion of polymorphisms appears as heterozygous. This may have been caused by primers used for viral genome amplification overlapping with polymorphisms present in the sample, as reported here: https://primer-monitor.neb.com/lineages.

To do:

• add fastqc [DONE]
• add multiqc
• mark duplicates [DONE]
• filter VCF using QUAL and DP [DONE]
• annotate VCF with snpEff [DONE]
• add unit tests for new rules
• update Dockerfile and Docker image with new environments
• use os.path.join for paths in Snakefile
• use snakemake wrappers
• add parallel-fastq-dump to fetch fastq by SRR id