PhaseByTransmission reassigns genotype code that conflicts with the allelic depths in the VCF

jlkuanjlkuan SingaporeMember

I encountered a problem running PhaseByTransmission (PBT) module on female proband with both healthy parents. Prior to running the module, the correct genotype was assigned to the Father but after that, the genotype was re-coded wrongly, supported by the allelic depths for ref and alt in the VCF. So would like to know why is this happening and how to prevent this error.

The PBT command used:
java -Xmx2g -jar GenomeAnalysisTK.jar -R human_g1k_v37.fa -T PhaseByTransmission -V trio_combined_haplotypeCaller.vcf -ped trio.ped -o pbt_trio_combined_haplotypeCaller.vcf

The following are the lines from the VCFs showing the variant record that has problem with the genotype coding. The sample columns are Proband, Mother and Father from left to right in the VCFs:

Before PBT (trio_combined_haplotypeCaller.vcf):
X 153296071 . GGGGGGCTGGTGGGGTCCTCGGAGCTCTCGGGCTCAGGTGGAGGT G 257.13 PASS AC=1;AF=0.167;AN=6;BaseQRankSum=-1.311e+00;ClippingRankSum=0.495;DP=73;FS=2.782;GQ_MEAN=138.00;GQ_STDDEV=132.68;MLEAC=1;MLEAF=0.167;MQ=60.00;MQ0=0;MQRankSum=1.31;NCC=0;QD=0.70;ReadPosRankSum=0.990;SOR=1.911;VQSLOD=1.51;culprit=ReadPosRankSum;set=variant2 GT:AD:DP:GQ:PL 0/1:17,8:25:99:288,0,4362 0/0:32,0:32:90:0,90,1350 0/0:13,0:13:36:0,36,540

After PBT (pbt_trio_combined_haplotypeCaller.vcf):
X 153296071 . GGGGGGCTGGTGGGGTCCTCGGAGCTCTCGGGCTCAGGTGGAGGT G 257.13 PASS AC=1;AF=0.167;AN=6;BaseQRankSum=-1.311e+00;ClippingRankSum=0.495;DP=73;FS=2.782;GQ_MEAN=138.00;GQ_STDDEV=132.68;MLEAC=1;MLEAF=0.167;MQ=60.00;MQ0=0;MQRankSum=1.31;NCC=0;QD=0.70;ReadPosRankSum=0.990;SOR=1.911;VQSLOD=1.51;culprit=ReadPosRankSum;set=variant2 GT:AD:DP:GQ:PL:TP 0|1:17,8:25:99:288,0,4362:43 0|0:32,0:32:90:0,90,1350:43 1|0:13,0:13:0:0,36,540:43

PED file content:
fam father 0 0 1 1
fam mother 0 0 2 1
fam proband father mother 2 2

Please advise.

Best Answers

  • jlkuanjlkuan Singapore
    Accepted Answer

    Hi Sheila,

    The problem was fixed after changing the default value for --DeNovoPrior from 1e-8 to 1e-6.

    Thanks,
    JL

Answers

  • jlkuanjlkuan SingaporeMember
    edited October 2016

    Hi Sheila,

    Attached is a snapshot of the BAM coverage for the father in the PDF.

    -JL-

  • jlkuanjlkuan SingaporeMember

    Hi Sheila,

    Does this mean the 11 reads supporting the Father as hom-ref are all false positive? This is because we normally filter the variants based on genotype code and it would be really helpful to understand how PBT decides the genotype despite conflicting the allelic depths.

    JL

  • SheilaSheila Broad InstituteMember, Broadie, Moderator admin

    @jlkuan
    Hi JL,

    Can you confirm you are using the latest version of GATK? Also, can you post the bamout file for the father?

    Thanks,
    Sheila

    P.S. Please post about 100 bases before and after the site of interest. Also, can you confirm the mapping qualities and base qualities are good at the site?

  • jlkuanjlkuan SingaporeMember
    edited November 2016

    Hi Sheila,

    I used GenomeAnalysisTK-3.3-0/GenomeAnalysisTK.jar for the HaplotypeCalling.

    Attached is the bamout for the father and also the details for all the 10 reads.

    Not sure how to check the mapping and base qualities because there was no variant called at this site during HaplotypeCalling stage for the father.

    Thanks,
    JL

  • SheilaSheila Broad InstituteMember, Broadie, Moderator admin

    @jlkuan
    Hi JL,

    For the mapping qualities, you can click on the read and have a look at the box that pops up. It will show the MQ of the read. If you click on the exact position, the box will also show the base quality at the position.

    Also, can you try running with the latest version? You can simply run on a small interval surrounding the site.

    Thanks,
    Sheila

  • jlkuanjlkuan SingaporeMember
    Accepted Answer

    Hi Sheila,

    The problem was fixed after changing the default value for --DeNovoPrior from 1e-8 to 1e-6.

    Thanks,
    JL

  • bhanuGandhambhanuGandham Member, Administrator, Broadie, Moderator admin

    Thank you for the update @jlkuan. This will help other users.

Sign In or Register to comment.