I'm currently performing some initial analyses on targeted Illumina sequencing data generated with DNA from diseased tissue. My hypothesis is that one particular gene might contain somatic mutations in this disease. I'm currently running Mutect2 in tumor-only mode, using default settings.

As I have only found very few mutations (this result might well be plausible), I'd like to adjust the Mutect2 cutoffs to increase the sensitivity and see whether the result changes. For instance, I'd like to reduce VAF, depth of coverage and QUAL. Unfortunately, I can't work out which of the many arguments have to be modified and how: https://gatk.broadinstitute.org/hc/en-us/articles/360037593851-Mutect2

Can you help me out? Thanks!

gatk Mutect2 \
-R $fastaFile \
-I $inputFile \
--germline-resource $germlineResource \
-O $resultDir/$sampleName.Mutect.vcf.gz

1 Answer 1


Mutect2 uses a statistical model based on the available reads. Unlike with some other tools, there is not threshold based on depth or VAF or QUAL. The main statistic that does govern emission is the LOD (Log ODds) which is the log10 of the odds ("ratio") that the variation being inspected is a tumor relative to the probability that it is a technical artifact. The default threshold for this is 3.0 (corresponding to a odds ratio of 1:1000).

Perhaps you'd like to see lower probability results, in which case you should lower the threshold:

Using --tumor-lod-to-emit 2.0 should give you many more results, many of them might be wrong. The LOD isn't well calibrated, so I'd take it with a grain of salt....validate the results you get.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.