To celebrate the release of GATK 4.0, we are giving away free credits for running the GATK4 Best Practices pipelines in FireCloud, our secure online analysis portal. It’s first come first serve, so sign up now to claim your free credits worth $250. Sponsored by Google Cloud. Learn more at https://software.broadinstitute.org/firecloud/documentation/freecredits

Filtering MuTect2 output for depth of coverage

kmegqkmegq BroadMember, Broadie

Dear GATK team,

I am writing with a question about depth of coverage in the output of MuTect2 (GATK3 and GATK4). I would like to filter my somatic calls to keep only calls with sufficient depth of coverage in both the tumor and the normal. However, I am unsure of the optimal approach.

The first approach I took was to add together the AD annotations for both the ref and alt alleles. However, my understanding from reading the forums is that the AD includes filtered reads. Because of this, I was concerned that some sites might actually have filtered depths lower than what was shown, and I wouldn't be able to filter those sites adequately.

The next approach I took is to use the DepthOfCoverage tool to annotate the filtered coverage at each site in the input bam file, and then look up the depths of each variant. This works, but does not take into account the fact that MuTect2 does local realignments which may change the depth of coverage around a variant.

So, I next looked into using the DepthOfCoverage tool on the MuTect2 bamout file. However, the depth of the normal sample in particular seemed really low compared to the input bam. Is there downsampling going on?

Do you have any thoughts on which of these annotations would be best to use to capture the depth for determination of whether we can confidently make a variant call at a particular position?

Is filtering for depth not something that you are expecting users to do, because the information on whether we can make a confident call is captured in the TLOD score, or another annotation?

Thank you for your help!

Best,
Kate

Tagged:

Comments

  • SheilaSheila Broad InstituteMember, Broadie, Moderator

    @kmegq
    Hi Kate,

    Is filtering for depth not something that you are expecting users to do, because the information on whether we can make a confident call is captured in the TLOD score, or another annotation?

    That is correct :smiley: We don't recommend users filter on depth because the tool takes into account many other factors that can influence the confidence in the calls. Have a look at this thread for more information (it is still relevant to Mutect2 even though the threads talks about HaplotypeCaller).

    -Sheila

Sign In or Register to comment.