Project Team 8 for HackSeq 2016 in Vancouver BC, Canada. The repository contains two projects.

Calling somatic mutation from tumor tissues only is challenge not only because you do not have a control to facilitate filtering out germline variants but it is difficult to differentiate low frequency somatic mutation from sequence noise/errors. In this study, we investigate whether we can leverage phasing information from reads to help differentiate somatic variants from germline alterations and sequencing errors.

This repository codebase dependes on 10xGenomics' longranger toolset. Download and install longranger. It also depends on the linked-reads data from 10xGenomics.

Get the counts of alt/ref (hap1, hap2, unphased, chrom, pos, from VCF file.

python count.py [--bed=<bed>] <ref_path> <vcf_path> <bam_path> <output_csv_path>

return value: Writes to disk a CSV file (given by output_csv_path) with columns : alt,chrom,filter,h1_alt,h1_ref,h2_alt,h2_ref,in_bed,pos,ref,un_alt,un_ref

Test run the somatic test on phased allele count data.

python somatic_test <count_file> <result_file>