Genotyping samples at a predetermined list of locations

Question

[i.sudbery]

Tools like HaplotypeCaller look at a BAM file and locate places where it differs from the reference. Thats great for variant discovery, but not so great if you want to genotype a sample, because distinguish between being homozygous reference and not having sufficient data to call a variant. That is, its not safe to assume a sample is homozygous reference just because HaplotypeCaller doesn't call a SNP.

Is there an obvious way to provide a redetermined set of locations (say dbSNP) and return, say, the properly quality adjusted log-likelihood of each genotype at each location?

[WouterDeCoster]

Not sure if that fully solves your question, but Freebayes can take a VCF as input for force-calling of those positions.

Answer 1

[bari.ballew]

You can force-call over a set of alleles with HaplotypeCaller. Look at the flags --force-call-filtered-alleles and --alleles.