search

dnanexus-rnd / GLnexus

Scalable gVCF merging and joint variant calling for population sequencing projects
Apache License 2.0
137 stars 36 forks source link

Incorrect variant calls #265

Closed gevro closed 2 years ago

gevro commented 2 years ago

Hi, For GLnexus version:

GLnexusVersion=v1.3.1-0-g0e1c9c9

GLnexusConfigName=gatk

GLnexusConfigCRC32C=1263441421

I'm getting missing variant calls in the jointly called file, even though the original gVCF of the input sample had a variant call.

Joint called VCF for locus -> Sample cmh003679-02 has ./. as the genotype with 'II' as the reason for the missing call, which means the input gVCF was not genotyped.

#CHROM    POS  ID   REF  ALT  QUAL FILTER    INFO FORMAT    cmh003679-01     cmh003679-02   cmh003679-03   cmh003679-04
chr5 77298807  chr5_77298807_CT_C  CT   C    61   .    AF=0.25;AQ=61     GT:DP:AD:SB:GQ:PL:RNC    0/1:19:6,13:.,.,.,.:61:61,0,72:..     ./.:5:5,0:.,.,.,.:0:.:II 0/0:30:30,0:.,.,.,.:50:.:..     0/1:13:6,7:.,.,.,.:59:58,0,99:..

However, that sample did have a genotype call at that locus...

#CHROM    POS  ID   REF  ALT  QUAL FILTER    INFO FORMAT    cmh003679-02
chr5 77298808  .    T    <*>  0    .    END=77298808   GT:GQ:MIN_DP:PL     0/0:15:5:0,15,149

This is probably a bug in GLnexus.

gevro commented 2 years ago

Actually perhaps this is due to the configuration having a minimum GQ. So I'm guessing that GLnexus treats variants that did not pass the filter as completely not genotyped when joining.