Differences between Mutect2 and MuTect2 single-sample calling
We are currently doing somatic mutation calling on two sets of cell samples. The first set is CD4+ cells and the other is CD8+ cells. We have both samples from the same patients. We are doing paired and single-sample calling on these samples. We have done the tumor-normal-calling in both directions (CD8+ as tumor, CD4+ as normal and the other way around) using Mutect2 and the results have been great. We have verified variants on IGV and for some of the samples with amplicon-sequencing.
However, the single-sample-calling has caused some issues with Mutect2. For some reason Mutect2 doesn’t emit some of the sites that have been called with tumor-normal-calling. The percentage it misses varies between samples but it can be up to half of the variants found with tumor-normal-calling. When we apply MuTect2 (GATK version 3.7) we get a lot more variant calls and can find all of the variants found with tumor-normal-calling from the initial output-file of MuTect2. Even by lowering the LOD-parameters down to 0.05 Mutect2 didn’t emit the sites that we are interested in. Could you briefly explain the differences between Mutect2 and MuTect2 that might have an impact on this issue? Which one would you recommend using?
Here is the Mutect2 command that I used:
gatk Mutect2 \ -R $REF \ -I $sample \ -tumor $sample_name \ -L $intervals \ --panel-of-normals $panel_of_normals \ -O $DIR/$outfile \ --germline-resource /resources/af-only-gnomad.hg38.no_chr.vcf \ -A StrandBiasBySample \ --initial-tumor-lod 0.05 \ --normal-lod 0.05 \ --tumor-lod-to-emit 0.05