CombineVariants and allele frequency

lalirlalir CanadaMember

Hello everyone,
I recenltly ran -T CombineVariants on a set of samples and I noticed that the allele frequency (AF) in the info field changed. In the individual VCF, AF=0.3 for the particular variant I was looking at, however, after I combined multiple VCFs, AF=0.5.... however the variant was only seen in the one sample, so how could the frequency change?
Any insight is appreciated.

Thank you,



  • SheilaSheila Broad InstituteMember, Broadie, Moderator admin


    Hi Ricky,

    Can you please post the records from the individual vcf and the CombineVariants vcf?


  • lalirlalir CanadaMember
    edited December 2014

    Sure, here is the example I was talking about: @Sheila

    ####SNP in individual sample####
    chr22 38509628 rs141777179 C T 86 PASS AF=0.393443;AO=24;DP=61;FAO=24;FDP=61;FR=.;FRO=37;FSAF=1;FSAR=23;FSRF=16;FSRR=21;FWDB=-0.0770567;FXX=0;HRUN=1;HS;LEN=1;MLLD=41.0391;QD=5.63915;RBI=0.206185;REFB=0.00282476;REVB=0.191245;RO=37;SAF=1;SAR=23;SRF=16;SRR=21;SSEN=0;SSEP=0;SSSB=-0.663883;STB=0.898401;STBP=0;TYPE=snp;VARB=-0.00219866;OID=rs141777179;OPOS=38509628;OREF=C;OALT=T;OMAPALT=T GT:GQ:DP:FDP:RO:FRO:AO:FAO:AF:SAR:SAF:SRF:SRR:FSAR:FSAF:FSRF:FSRR 0/1:85:61:61:37:37:24:24:0.393443:23:1:16:21:23:1:16:21

    ####SNP in ALL samples####
    chr22 38509628 rs141777179 C T 86 PASS AC=1;AF=0.500;AN=2;AO=24;DP=61;FAO=24;FDP=61;FR=.;FRO=37;FSAF=1;FSAR=23;FSRF=16;FSRR=21;FWDB=-0.0770567;FXX=0;HRUN=1;LEN=1;MLLD=41.0391;OALT=T;OID=.;OMAPALT=T;OPOS=38509628;OREF=C;QD=5.63915;RBI=0.206185;REFB=0.00282476;REVB=0.191245;RO=37;SAF=1;SAR=23;SRF=16;SRR=21;SSEN=0;SSEP=0;SSSB=-0.663883;STB=0.898401;STBP=0;TYPE=snp;VARB=-0.00219866;set=variant GT:AF:AO:DP:FAO:FDP:FRO:FSAF:FSAR:FSRF:FSRR:GQ:RO:SAF:SAR:SRF:SRR 0/1:0.393443:24:61:24:61:37:1:23:16:21:85:37:1:23:16:21 ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./.

    variants were called using torrent variant caller, that is why some of the annotations are different. As you can see, this variant is present in only the one sample. The AF however is showing as 0.5, shouldn't it be the same as in the individual vcf file?

  • Geraldine_VdAuweraGeraldine_VdAuwera Cambridge, MAMember, Administrator, Broadie admin

    @lalir, this is because GATK calculates AF differently than other callers. The AF reported by GATK is the expected AF for a particular genotype (so for a diploid het, the AF will always be reported as 0.5), whereas the torrent caller probably reported the effective AF observed in the data. GATK currently does not include an annotation to get the empirical AF, but you can calculate it using the reported allele depths (AD).

Sign In or Register to comment.