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.

Mutect2 repeatedly not detecting somatic variant IDH2 R172K, with solid read support and 5% AF

mack812mack812 SpainMember
edited October 14 in Ask the GATK team


I am currently validating a GATK analysis pipeline applied to data from amplicons and deep coverage (>1000x) for detection of somatic variants at hotspot sites in cancer samples. As part of this validation I have included reference samples from a commercial provider, which harbour many relevant variants at defined allelic frequencies. The result of the validation so far is quite satisfactory overall, especially for a limit of detection of 5% (FYI 1% AF-LOD is worse, and detection is highly dependent on depth of coverage, requiring more supporting reads than I expected). However, there is a single variant (IDH2 R172K) that goes consistently undetected by Mutect2, in two independent runs of this sample. The variant is present at 5% allelic frequency and was covered at 2000-3000x (100-150 supporting reads) and therefore was not expected at all to cause any troubles to Mutect2. The variant can be easily seen in IGV and has very deep coverage in the bamout (hence it is not a problem of an active region not being triggered at this site or of reads being filtered out by the tool).

To make this issue even more difficult to understand, another variant less than 15 bp away was correctly detected. Moreover, this same variant (IDH2 p.R172K) was detected in a different sample at an even lower AF (1%). So, it seems that Mutect2 is unable to detect the variant IDH2 p.R172K in the specific sequence context present at this sample (as said before, this happened twice: in two independent runs).

I am using GATK and the following Mutect2 command:

gatk --java-options "-Xmx4000m" \
  Mutect2 \
    -R $ref_fasta \
    -I $inpath/$filename \
    -germline-resource $gnomeAD_af \
    -L $interval_hotspots \
    --max-reads-per-alignment-start 0 \
    --disable-read-filter MateOnSameContigOrNoMappedMateReadFilter \
    --disable-read-filter NotDuplicateReadFilter \
    -O $vcf_outpath/${basename}.unfiltered.vcf \
    -bamout $vcf_outpath/${basename}.bamout.bam

I am concerned that this could be happening at other regions, potentially increasing the rate of false negatives, so I would appreciate if you could help me identify the underlying problem here.


Sign In or Register to comment.