Heads up:
We’re moving the GATK website, docs and forum to a new platform. Read the full story and breakdown of key changes on this blog.
If you happen to see a question you know the answer to, please do chime in and help your fellow community members. We encourage our fourm members to be more involved, jump in and help out your fellow researchers with their questions. GATK forum is a community forum and helping each other with using GATK tools and research is the cornerstone of our success as a genomics research community.We appreciate your help!

Test-drive the GATK tools and Best Practices pipelines on Terra

Check out this blog post to learn how you can get started with GATK and try out the pipelines in preconfigured workspaces (with a user-friendly interface!) without having to install anything.
We will be out of the office for a Broad Institute event from Dec 10th to Dec 11th 2019. We will be back to monitor the GATK forum on Dec 12th 2019. In the meantime we encourage you to help out other community members with their queries.
Thank you for your patience!

Calibration of QUAL field emitted by the Unified Genotyper

LaurentLaurent Member, Broadie ✭✭


I am looking at the distribution of quality scores (QUAL column of VCF) emitted by the Unified Genotyper v1.4 (old version, sorry :)) for SNPs called on 769 samples with 12x coverage HiSeq data. The median quality score is at ~Q750 and the mean at ~Q28000. I was just wondering if you could comment on how well calibrated these quality scores are and whether/how the number of samples and/or coverage would affect them?

Thanks in advance,

Best Answer


  • Geraldine_VdAuweraGeraldine_VdAuwera Cambridge, MAMember, Administrator, Broadie admin

    Hi Laurent,

    I'm not quite sure what you mean by calibration for the variant quals? The QUAL will indeed depend on coverage and number of samples, since both can increase or decrease confidence in a given call according to the calling model. Basically it is a question of how many observations are available from which to make a call. If you want to known exactly what is the mathematical relationship you'll need to look at the code of the model calculations for the version that you are using.

  • LaurentLaurent Member, Broadie ✭✭

    Hi Geraldine,

    Thanks for your answer. What I mean is that QUAL is expressed as a phred score (if I'm not mistaking) and so Q28000 is a rather astronomical number then. I'm just trying to get a sense of how to interpret these values and whether I should really look at them in a relative way (i.e. in the same calling experiment, a site with higher quality has a higher posterior) or if they can be interpreted in an absolute way too.


  • LaurentLaurent Member, Broadie ✭✭

    Thanks Geraldine,

    This was helpful in clarifying my issue!


Sign In or Register to comment.