Het site 0/1 becoming ./. in the pVCF

dnanexus-rnd / GLnexus

Scalable gVCF merging and joint variant calling for population sequencing projects

Apache License 2.0

142 stars 37 forks source link

Het site 0/1 becoming ./. in the pVCF #215

Closed ghost closed 4 years ago

ghost commented 4 years ago

Hello,

I have a case where at some position a reference sample has the GT 0/1

Chrom_3 11583331    .   G   T,<*>   35.2    PASS    .   GT:GQ:DP:AD:VAF:PL  0/1:35:281:134,140,0:0.498221,0:35,0,55,990,990,990

And 1/1 in another sample

Chrom_3 11583331    .   G   T,<*>   33.7    PASS    .   GT:GQ:DP:AD:VAF:PL  1/1:20:44:0,44,0:1,0:33,19,0,990,990,990

However, in the pVCF

Chrom_3 11583331    Chrom_3_11583331_G_T    G   T   33  .   AF=0.5;AQ=33    GT:DP:AD:GQ:PL:RNC  ./.:.:.:.:0,0,0:MM  1/1:44:0,44:16:33,19,0:..

The genotype of the first sample has moved from 0/1 to ./.

Is it an intended feature?

Thank you

mlin commented 4 years ago

There are some complex situations where that can happen legitimately (discussed on Reading GLnexus pVCFs), however this doesn't appear to be one of them based on the information provided. The RNC field "MM" (M for Missing) is the genotyper claiming it did not see the first record at all. Is this an isolated discrepancy or is the sample exhibiting more general problems? Any overlapping gVCF records or anything otherwise unusual in the vicinity?

ghost commented 4 years ago

yes it was more general ... However your comment about "Missing" made me wonder if I could reproduce it, and I can't. I deleted the pVCF and generated it again. And this time, no problem.

I have no idea of what happened, but apparently something went wrong during the first run.