"Raw" VCF FILTER field, previous GATK builds vs Current

bwubbbwubb Posts: 48Member
edited September 2012 in Ask the GATK team

Hello, Did the UnifiedGenotyper of previous builds use to place "PASS" in the filter fields of vcf files? I have re-ran some data using current best practices, including the HaplotypeCaller. My first snps.indels.raw.vcf file has all . in the FILTER field, which I remember was not a good thing. Im compared the vcf produced at this step with my previous UnifiedGenotyper counterpart and the FILTER field was populated with "PASS". I am bit concerned that is there were no PASS then no LowQual filter applied by the caller either.

Besides using HaplotypeCaller, the only other argument I had changed was returning the -stand_emit_conf to the default. Is this perhaps the root of my concern? Does having both stand_call_conf and stand_emit_conf at defaults ie. equal not apply any FILTER?

Thank you.

Best Answer


  • ebanksebanks Posts: 684GATK Developer mod

    The behavior should not have changed. "." in the filter field is not bad (it means that no filtering was applied). Please see the VCF spec for more information.

    Eric Banks, PhD -- Senior Group Leader, MPG Analysis, Broad Institute of Harvard and MIT

  • bwubbbwubb Posts: 48Member
    edited September 2012

    Understood. I guess Im just curious how I applied a filter previously when running UnifiedGenotyper for the first time. I apologize if this is a bit frivolous, but its driving me a bit crazy.

    My current vcf header has ##FILTER=<ID=LowQual,Description="Low quality">, but I do not see any LowQual or PASS, so I am inclined to believe this was not applied (as you suggested) and I would like to change that, especially if it can be applied automatically during initial variant calling (before I apply other hard filters). I will have to change my -stand_emit_conf again. Im guessing that is applied to the QUAL? Thank you.

    Post edited by bwubb on
Sign In or Register to comment.