missing DNMs from Strelka call

[weizhu365]

I have tried to use slivar to call DNM from Strelka prediction:

slivar expr --vcf output/slivar/strelka_t0334.tmp0.vcf.gz \ --ped ped_files/t0334.ped --pass-only \ --out-vcf output/slivar/strelka_t0334.dnm.vcf \ --trio "denovo:kid.het && mom.hom_ref && dad.hom_ref \ && kid.AB > 0.25 && kid.AB < 0.75 \ && (mom.AD[1] + dad.AD[1]) <= 5 \ && kid.GQ >= 20 && mom.GQ >= 20 && dad.GQ >= 20 \ && kid.DP >= 30 && mom.DP >= 30 && dad.DP >= 30"

And several variants were failed in the slivar filtering, for example: chr1 243356830 . GA G 659 PASS CIGAR=1M1D;RU=A;REFREP=10;IDREP=9;MQ=60 GT:GQ:GQX:DPI:AD:ADF:ADR:FT:PL 0/0:136:136:57:47,0:24,0:23,0:PASS:0,139,999 0/0:184:184:75:63,0:33,0:30,0:PASS:0,187,999 0/1:435:27:70:28,38:19,20:9,18:PASS:670,0,432

In my manual inspection, I think it should be "PASS". Could you tell me why it was failed to pass the filters?

Thanks,

Wei

[brentp]

Hi Wei, could you share a VCF with only this variant and the VCF header along with the ped file for these 3 samples?

[weizhu365]

Many thanks for your prompt reply. Here are the input files you requested:

strelka_issue.zip

[brentp]

Hi Wei, when I run your variant, I see this warning:

slivar] javascript error. this can some times happen when a field is missing.
error from duktape: unknown attribute:DP for expression:kid.het && mom.hom_ref && dad.hom_ref && kid.AB > 0.25 && kid.AB < 0.75 && (mom.AD[1] + dad.AD[1]) <= 5 && kid.GQ >= 20 && mom.GQ >= 20 && dad.GQ >= 20 && kid.DP >= 30 && mom.DP >= 30 && dad.DP >= 30

note unknown attribute:DP.

slivar tries to warn on these cases as sometimes a field is missing from a few VCFs, but the expression can not pass without that information. I change your expression to:

denovo:kid.het && mom.hom_ref && dad.hom_ref && kid.AB > 0.25 && kid.AB < 0.75 && (mom.AD[1] + dad.AD[1]) <= 5 && kid.GQ >= 20 && mom.GQ >= 20 && dad.GQ >= 20 && kid.AD[0] + kid.AD[1] >= 30 && mom.AD[0] + mom.AD[1] >= 30 && dad.AD[0] + dad.AD[1] >= 30

and it passes as expected. Hope this helps. -Brent

[weizhu365]

Thank you for the solution. I found that some variants in the same VCF file had passed the filters, which having DP attributes. It is weird that not all variants have the same attributes in the Strelka output.

[brentp]

From the header of your VCF, it looks like it uses DPI for indels and DP for other variants.

[weizhu365]

That makes sense. Best,