VarScan: somatic mutations in heterozygous sites?

[ElenaN]

Does VarScan2 find somatic mutations that arose in loci where norm was heterozygous? As far as I can tell, it classifies somatic changes as "Somatic" (but that seems to only include homozygous normal) and "loss of heterozygosity", where normal was hetereozygous, and tumor is homozygous". What happens if normal was heterozygous, and tumor has 3 alleles (one novel)? I don't seem to be able to find such sites in VarScan output.
Thank you,
Elena

[dkoboldt]

Are there situations where true, validated somatic mutations arise at the position of heterozygous SNPs? My expectation is that most such situations would represent artifactual calls, not real events.

In your example, the normal would be called heterozygous, and the tumor would be called heterozygous (if a variant allele met calling criteria). The most-supported variant allele in tumor would be taken as the consensus. If it's the same as the normal variant allele, the call will be Germline. If it's a third allele, VarScan will call it "Unknown" because this is an unexpected situation. Running VarScan somatic with the --validation 1 parameter will tell you.

You could also investigate by running VarScan mpileup2cns instead. Feel free to e-mail me the pileup outputs if you'd like to discuss further.