Germline variant calling using tumor-normal pairs in GATK

[ty23991]

Hi
i have tumor normal pairs and am interested in calling germline variants. The GATK UnifiedGenotyper does seem to call the germline variants as it does not require normal/tumor input arguments.

I don't think there are any options in GATK.

Apart from varscan2, what other tools can be used for germline calling based on tumor-normal pairs ?

And which one performs the best ?

[Bukowski]

The GATK based option is MuTect: https://www.broadinstitute.org/cancer/cga/mutect

[ty23991]

Mutect is good for somatic variant calls, not the germline.

[Bukowski]

Er it calls the tumour and normal precisely to separate the somatic from the germline variants.. that's the whole point of doing tumour/normal calling right?

If you only want to call the germline mutations - just use GATK anyway and remove your variants identified as somatic elsewhere. I'm surprised MuTect doesn't output somatic and germline.

[ty23991]

Yes. I agree. But I m not sure what filters can I use for the germline variants .

[ty23991]

Er it calls the tumour and normal precisely to separate the somatic from the germline variants.. that's the whole point of doing tumour/normal calling right?

If you only want to call the germline mutations - just use GATK anyway and remove your variants identified as somatic elsewhere. I'm surprised MuTect doesn't output somatic and germline.

I checked my unfiltered vcf files but could not find any with normal and tumor both with mutant genotype. So mutant does not seem to output germline at all.

Mutect also uses a filtering criterion called "germline risk variant" but that's not the actual germline variant call.