A false-positive filter for variants called from massively parallel sequencing

Update (5/30/2017): This repo is no longer maintained. Use the fpfilter.pl at https://github.com/genome/fpfilter-tool instead.

Download the fpfilter.pl script, and view the detailed usage manual:

curl -LO https://github.com/ckandoth/variant-filter/archive/master.zip; unzip master.zip
perl variant-filter-master/fpfilter.pl --help

Given snvs.vcf a single sample VCF listing SNVs, generate a list of variant loci in a subfolder:

mkdir fpfilter
perl -ane 'print join("\t",@F[0,1,1])."\n" unless(m/^#/)' snvs.vcf > fpfilter/snvs.var

Given tumor.bam, a BAM file containing the SNVs, and its reference FASTA grch37.fa, run bam-readcount:

bam-readcount -q1 -b15 -w1 -l fpfilter/snvs.var -f grch37.fa tumor.bam > fpfilter/snvs.readcount

Run the fpfilter.pl script using the readcounts as an input:

perl fpfilter.pl --var-file snvs.vcf --readcount-file fpfilter/snvs.readcount --output-file fpfilter/snvs.fpfilter

The last two columns of the resulting file snvs.fpfilter contains a VCF friendly FILTER code, and an extra column with details.

The original fpfilter.pl script for filtering Single Nucleotide Variants (SNVs) was written by Dan Koboldt for VarScan, and Dave Larson for SomaticSniper. The latest version of the script is maintained by Dan, Dave, and other developers at the The Genome Institute, as a Perl module named FilterFalsePositives.pm. They even wrote a module named FilterFalseIndels.pm for small indels. However, those modules are not very portable, and have dependencies on their internal codebase. So this repository was my attempt to remove those dependencies, make it easier to use, and add other improvements like VCF support. More recently, there have been forks of this repo, and this one is what I recommend using, which also does non-complex indels.

Cyriac Kandoth ([email protected])
Dan Koboldt, Analysis Manager, Human Genetics Group, The Genome Institute
David Larson, Director of Bioinformatics, The Genome Institute