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!

Mutect2 dont list Variant

EADGEADG KielMember ✭✭✭

Iam running mutect2 on capture data, with tumor normal. The data flows the "classic" way on a local wdl-pipeline with bwa =>merge=>markDup=>recal=>mutect2 (3.6).

I see the variant clearly in IGV in both tumor and normal sample and even in the bamout i created for it (see screenshot). But it dont get listed in the output vcf.

I know that mutect2 apply some hard filters to the variant, but I should see the variant in the output marked with an alt_allele_in_normal ?





Best Answer


  • shleeshlee CambridgeMember, Broadie ✭✭✭✭✭

    Hi @EADG,

    I see the variant clearly in IGV in both tumor and normal sample

    MuTect2 is a somatic caller that only calls on differences in the tumor from the normal. However, as you say, the call should be present with an annotation that filters it. Your screenshots, are these the raw BAMs or the BAMOUTs from a MuTect2 run?

  • EADGEADG KielMember ✭✭✭

    Hi @shlee,
    thank you for you answer...the screenshots are showing in order tumor.raw.bam, normal.raw.bam and tumor-normal.bamout.bam.

  • shleeshlee CambridgeMember, Broadie ✭✭✭✭✭


    The BAMOUT needs to by grouped by sample, colored by HC tag and sorted by HC tag. Also, can you show the reference sequence and genomic location for this locus? You've previously done a helpful IGV coloring at the end of this thread for a MuTect2 BAMOUT.

    In general, we need more information than what you've so far provided to effectively answer questions such as why a variant wasn't called. Please help us help you more effectively by provided the needed information up front so we can minimize this back and forth. We are a small team. Thanks.

  • EADGEADG KielMember ✭✭✭

    sry:(...i made a new screenshot and hope it will satisfy and also I will provide some helpful information (hope so :).
    I run the pipline in the broadinstitute/genomes-in-the-cloud 2.2.4-1469632282 Docker container with cromwell version 21, so its gatk 3.6.
    Furthermore i use scatter/ghatter and merging the resulting vcf´s. My command for mutect2 is:

        java -Xmx4g -jar /usr/gitc/GATK36.jar \
        -T MuTect2 \
        --dbsnp ${dbSNP_vcf} \
        -dt NONE \
        -R ${ref_fasta} \
        -I:normal ${input_normal_bam} \
        -I:tumor ${input_tumor_bam} \
        -L ${interval_list} \
        -o ${vcf_basename}.MT.vcf

    Say a what you want and I deliver more information :).

    Thank you for your help!


    Issue · Github
    by shlee

    Issue Number
    Last Updated
    Closed By
  • shleeshlee CambridgeMember, Broadie ✭✭✭✭✭

    Thanks for supplying a better screenshot. I hope there is an easy explanation for this odd phenomena @EADG. Would you mind generating a snippet of the data needed to recreate this image, i.e. your tumor and normal raw BAMs as well as your BAMOUT and callset? Instructions for sending us data are at http://gatkforums.broadinstitute.org/gatk/discussion/1894/how-do-i-submit-a-detailed-bug-report.

  • EADGEADG KielMember ✭✭✭

    Hi @shlee

    i manage to upload the Files to your server (gosh that 20 user limitation is quite an annoyance). The Folder is EADG-Mutect2.

    I have observe the same issue with an other sample which carries the same variant...Do you mean i it is useful when I check if i get the variant called with HC, since the relationship between HC and Mutect2 ?

    Thank you!

    Greetings EADG

  • EADGEADG KielMember ✭✭✭

    Hi @shlee,

    just to be shure I checked the position an noticed that all of them have rs-numbers and are included in dbSNP. I know you have plenty off other stuff to do...argh but you have a little status update for me ?

    Greetings EADG

  • shleeshlee CambridgeMember, Broadie ✭✭✭✭✭

    Hi @EADG,

    @Sheila has returned from her conference and she should be able to help you going forward, especially given your question requires in-depth probing. I've just been filling in for the team this last week since most are away.

  • SheilaSheila Broad InstituteMember, Broadie ✭✭✭✭✭


    I will have a look and get back to you soon.


  • Geraldine_VdAuweraGeraldine_VdAuwera Cambridge, MAMember, Administrator, Broadie admin
    As I recall M2 doesn't call obvious germline variants, or at least it's not supposed to, no? My impression was that we only output "germline_risk" records if they have a chance of being somatic and are borderline. If they're obviously germline, why bother? Would just make much bigger files for nothing... But I may have misunderstood the default behavior.
  • EADGEADG KielMember ✭✭✭

    Hi everbody and thank for your effort,
    if its like @Geraldine_VdAuwera said it is ok, but I need an "official" confirm, because I find nothing about it in the doc´s.
    Maybe a verbose-mode were good were mutect2 puts every think out even it is a obviously germline variant. Just to be shure everything is allright.

    Thanks so far!

  • Geraldine_VdAuweraGeraldine_VdAuwera Cambridge, MAMember, Administrator, Broadie admin

    @EADG We're checking on this with the M2 devs to get you that official confirmation, and document this more clearly as well. Sorry it's taking so long -- part of our team just got back from an international workshop so we're playing catch-up.

  • EADGEADG KielMember ✭✭✭


    Good to know :) Thanks for your efforts! Hope Belgium was nice!

Sign In or Register to comment.