Some questions on QD

Hello GATK team,

I want to apply this to some GATK processed data I obtained from a reliable source. I noticed that the QD values (in my data) do not match the procedure explained here. Some questions for you:

1) For multi-sample VCFs, how is the QD value computed by GATK when more than one sample is non hom-ref? For instance in my case I will have different QDs as QD=QUAL/AD, for different values of AD. If I compute my own QDs and I want to apply these filters, will it be reliable to do such filtering using my own ODs?

2) Is there a chance of getting a value different from QD=QUAL/AD if a specific argument is specified when executing the variant calling function (i.e HaplotypeCaller)?

Thank you,



Best Answer


  • cdenizcdeniz Member

    Hi @bhanuGandham,

    Yeah, I noticed the AD mistake. However, I've been calculating my QDs using QUAL/DP (as DP is the sum of the ADs). What I didn't know for sure was if DP = AD exclusively for non hom-ref ADs, therefore I assumed your initial equation might have been correct.

    Thank you, for the details upon the multi-sample QD.


Sign In or Register to comment.