If you happen to see a question you know the answer to, please do chime in and help your fellow community members. 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.

calling variants in multiple samples

i have sequenced two exomes from two affected individuals from the same family.
i have called their variants with the recommended UnifiedGenotyper protocol for each sample individually (was done a year and half ago..), and then did theintersection to find teh shared ones.
i was wondering, if you recommend calling them at one step, togther? is there any added value for doing so (in term of reducing false positives, accuracy etc..).


  • CarneiroCarneiro Charlestown, MAMember admin

    It is always better to call samples together to empower filtering. However in a 2 sample situation the gains may not be all that significant. It never hurts though.

  • galmorangalmoran Member

    thanks for your quick reply!
    so, what command should i be using?
    the analysis is only joined at the var call step (i.e, the processing of BAM files - local realignment, variant qaulity recalibration etc.. is done separately?). it should be noted they were sequenced in different experiments...
    i also see that since i've used GTAK last time, you have indtroduced the HaplotypeCaller. should i used it instead of UnifiedGenotyper?

    and last question, if you are willing to elaborate - what is exactly achieved when i call multiple samples stimultaneosly in one command?
    does it compare or filter them based on the comparison? (especially if they were not sequenced toegther)

  • Geraldine_VdAuweraGeraldine_VdAuwera Cambridge, MAMember, Administrator, Broadie admin

    Hi Moran,

    Multisample calling is addressed in the UnifiedGenotyper documentation and in many other posts on this forum. Please have a look through that and if you have any remaining questions we'll be happy to answer them.

    If your samples are diploid you should give the HaplotypeCaller a shot, yes. the indel results in particular will be much better. Multisample calling is achieved the same way with the HC as with the UG.

Sign In or Register to comment.