Per-allele filters?

jamesp

I'm doing analysis on a large multi-ancestry project (XX,XXX exomes) in which approximately 20% of variant sites are multi-allelic. Our sequencing/calling is done at the Broad. It seems that at these multi-allelic sites, only one filter is applied. For example, where there is a SNP and an indel, the SNP may get an indel filter.

Is it possible to disclose how the filter at a given site is chosen? (E.g., by severity, etc.) Is there a plan to allow each variant to have its own filter in the future? (Not sure if there is an internal Broad place to discuss this, since it may have to do more with the pipeline and how SNPs and indel calls are combined than the GATK per se.)

bhanuGandham

Hi @jamesp

Would you please send us the command you are using for variant filtering, the version of gatk and also some records in the vcf where you see this issue.

Once I get this information from you I can bring it up with our dev team.

Here is some info on filtering of germline short variants: https://software.broadinstitute.org/gatk/documentation/article?id=11070

jamesp

Unfortunately, I'm receiving this from the Broad core, so I don't know what commands they are using.

bhanuGandham

HI @jamesp

I guess I do not quite understand the question here. What is you input data? Were the variants already filtered before you received them? Can you ask the core what pipeline/tools they used?

bhanuGandham

We haven't heard from you in more than two business days. I will be closing this issue now. But please feel free to ask more questions if reqd.