We've moved!
This site is now read-only. You can find our new documentation site and support forum for posting questions here.
Be sure to read our welcome blog!

Calling multiple samples with UG

Hi all,

I was wondering if there is a recommended way how to call variants in trio data (affected children, healthy parents) using the Unified Genotyper. As I see it, there are three possible ways:

  1. Call each sample individually
  2. Call trios together
  3. Call all samples together

But which one would you recommend?

Best Answer


  • MaxMax Member

    Hi Geraldine,

    thanks a lot for your help!

    I'm aware of the HC and the new GVCF mode. The reason why I decided against the HC and still use the UG is basically due to this paper: http://www.ncbi.nlm.nih.gov/pubmed/24498629 in which they showed better validation rates for the UG for both, SNPs and Indels.

  • Geraldine_VdAuweraGeraldine_VdAuwera Cambridge, MAMember, Administrator, Broadie admin

    Ah, interesting paper -- thanks for bringing this to my attention, Max. One quick comment: that study used a much earlier version of HaplotypeCaller (2.4.9); sensitivity has massively improved since then. The authors make this point regarding versions and report that using a newer version (probably 2.7 considering typical publication timelines for BMC journals) showed notable improvements. I think their final recommendation (use cutting edge versions for research and more stable versions for clinical settings) is fair and makes a lot of sense -- although I would argue that HaplotypeCaller is getting to a mature enough stage now that it's worth anyone's while (including clinical folks) to evaluate the version 3.x tools. You wouldn't want to miss out on great features based on outdated info... For a sense of scale, the time elapsed since the version that was tested in the paper represents over half of the entire lifetime of the HaplotypeCaller to date, if I'm not mistaken. That's like evaluating a professional athlete based on his performance in Little League ;)

Sign In or Register to comment.