By the default configuration, it seems that all variants are somatic. I could not find the parameter which would regulate that.
That's right; to elaborate on Sheila's response, older versions emitted sites that were borderline, where there was some initial possibility that they might be somatic, but were then filtered out as most likely germline. The current version no longer emits these sites, and it's not trivial to restore that behavior because the program now rules them out much earlier in the process. There are some plans to provide some relevant information to help diagnose cases where expected variants may be missing. But in any case that behavior was never meant to provide proper germline variant calls.
Mutect2 is not a germline variant caller; its modeling algorithm is tuned specifically for somatic mutations. If you're looking for germline variation as well you'll need to use a germline caller like HaplotypeCaller in a separate run.
thanks for the fast reply. In the previous versions, correct me if I'm wrong, the output VCF contained both germline and somatic variants. Was that changed? Or maybe some parameter can prevent outputting of the germline variants. Here is the command line:
java -Xmx2048M -jar /opt/gatk-package-4.beta.3-local.jar Mutect2 --output Tumor_exomeSeq.vcf --tumorSampleName Tumor_exomeSeq --reference Homo_sapiens.GRCh38.dna.primary_assembly.fa --input tumor_recalibrated.bam --input normal_recalibrated.bam --normalSampleName Normal_exomeSeq --intervals 9.bed --dbsnp dbsnp_144.hg38.vcf
I am not aware of previous versions of Mutect outputting germline variants.
I am assuming you are asking about sites that are either in the PoN, germline resource or in the matched normal. For the sites in the PoN or germline resource, you can simply take a look at those in the resource files and compare them to the output of Mutect2. For the matched normal, it is best to run a germline variant caller like HaplotypeCaller. Note, FilterMutectCalls will also flag any sites called by Mutect2 that are in the PoN or germline resource or have a high LOD score in the normal.
Thanks a lot, @Geraldine_VdAuwera and @Sheila!