This is a test version of Biostars. For the public version, visit https://www.biostars.org.
GATK VariantFiltration multiple filters including genotype for multi-sample VCF

I have a multisample VCF and I want to filter it based on:

  1. The sequencing depth from the first sample (which is a germline sample) and
  2. The genotype of the first (and/or the third) samples (which are both germline samples)

I've been investigating gatk VariantFiltration for doing this, using -filter='vc.getGenotype("SAMPLE").getDP>=<N>' -filter-name="germline.depth" for the first sample sequencing depth, but I can't find a suitable expression involving the sample genotype that works for the second condition.

Ideally, an expression similar to this: -filter='<something>' -filter-name='something.else' would be used, as it puts the filter into the FILTER column of the VCF, and the unfiltered variants can be selected using gatk SelectVariants to pick those that haven't been marked as filtered.

Is there an expression like the depth one for the individual sample genotypes?

vcf gatk
• 2,502 views • link
written by graeme.thorn graeme.thorn

no tested, testing sample2 is HET

vc.getGenotype("SAMPLE1").getDP()>=12345 && vc.getGenotype("SAMPLE2").isHet()
• 0 views • link
written by Pierre Lindenbaum Pierre Lindenbaum

Thanks! A quick test has shown that variations of this (see the answer) are what I required. The GATK help pages don't seem that helpful.

• 0 views • link
written by graeme.thorn graeme.thorn

1 answer

As per @Pierre Lindenbaum's comment above, the correct filter for the genotype is

vc.getGenotype("SAMPLEn").isHomRef() to select the "0/0" genotype and !vc.getGenotype("SAMPLEn").isHet() && !vc.getGenotype("SAMPLEn").isHomVar() to select the "0/0" or "./." genotypes.

This covered all the use cases I needed.

• 0 views • link
written by graeme.thorn graeme.thorn

Log in to answer this question.

More posts like this