Combining Normals in multi-region genome sequencing

Question

[nicholasyounger38]

I'm analysing multi-region exome and genome sequencing data and alongside multiple tumour biopsy regions, we also have multiple adjacent normals from the same patient. I was wondering if it would be suitable to combine the normals to increase coverage of the 'normal' genome, and if so what the best way to do this would be?

If not, how would one choose which normal to use for mutation calling? I had considered looking for tumour contamination in the normals and simply picking the one with the least contamination, assuming there is any to begin with. But it seems a shame to waste additional normal samples.

Answer 1

[Kevin Blighe]

What about performing 'joint genotyping' for the purpose of coming up with a 'normal genome VCF (gVCF)'? - joint genotyping, as in, the process used by GATK. Even using bcftools mpileup, you can perform the pileup and variant calling from multiple input BAMs simultaneously. I think that either of these would be sufficient, to be honest.

[nicholasyounger38]

Thanks very much, I have asked around the bioinformaticians here also and joint genotyping seems to be the consensus approach. Much appreciated.