[possible-bug] GATK 3.4-46 CombineGVCFs is wrongly assigning * at -1 position of some intervals

We found that whenever the previous interval ends in a deletion, GATK is extending upstream 1 base the following interval and wrongly adds the * call to it. In most cases this will be a non-issue as you will filter downstream analysis with the same interval files and these lines won't be included. Still it looks like a bug the team might want to look into.

Also we are running CombineGVCFs in "BP_RESOLUTION" mode. Not sure if this makes a difference.

A couple of examples:

#CHROM  POS ID  REF ALT QUAL    FILTER  INFO
1   100330153   .   C   <NON_REF>   .   .   .
1   100330154   .   T   <NON_REF>   .   .   .
1   100330155   .   TC  T,<NON_REF> .   .   BaseQRankSum=-0.572;ClippingRankSum=0;DP=14876;MQ=60;MQRankSum=-1.067;ReadPosRankSum=0.572
1   100335944   .   T   *,<NON_REF> .   .   DP=8
1   100335945   .   T   <NON_REF>   .   .   .
1   100335946   .   T   <NON_REF>   .   .   .

#CHROM  POS ID  REF ALT QUAL    FILTER  INFO
1   215916683   .   AAGAG   A,AAG   9405.15 .   AC=4,81;AF=0.008547,0.173;AN=468;BaseQRankSum=-0.061;ClippingRankSum=0.1;DP=25502;FS=0;InbreedingCoeff=-0.2174;MLEAC=2,80;MLEAF=0.004274,0.171;MQ=60;MQRankSum=0.092;QD=1.25;ReadPosRankSum=-1.035;SOR=0.65
1   215916684   .   A   *,<NON_REF> .   .   DP=25594
1   215916685   .   G   *,<NON_REF> .   .   DP=25593
1   215916686   .   A   *,<NON_REF> .   .   DP=25128
1   215916687   .   G   *,<NON_REF> .   .   DP=25115
1   215931925   .   T   *,<NON_REF> .   .   DP=65
1   215931926   .   T   <NON_REF>   .   .   .
1   215931927   .   T   <NON_REF>   .   .   .

Relevant lines in the interval file:

1   100329930   100330155   AGL 0   +
1   100335944   100336147   AGL 0   +
1   215916507   215916687   USH2A   0   -
1   215931925   215932104   USH2A   0   -

Thanks,
Carlos

Tagged:

Answers

  • SheilaSheila Broad InstituteMember, Broadie, Moderator admin

    @CarlosBorroto
    Hi Carlos,

    I am not sure I follow what you are saying. Do you mean position 1:100335944 does not have a deletion in the GVCF, but in the Combined GVCF, it shows a deletion (*)? If so, then that is a bug and I will need you to submit some test files.

    Thanks,
    Sheila

  • CarlosBorrotoCarlosBorroto Member ✭✭

    Hi @Sheila,

    Yes, in the samples above positions 1:100335944 and 1: 215931925 are called as deletion but I don't think they are. I will collect test files and upload them to the FTP.

    --Carlos

Sign In or Register to comment.