Why are there too many mutations in controls, but not in tumour?

Question

[MAPK]

I have been studying different cancer samples and comparing them with the germline controls for driver mutations. As a result I have found quite of few somatic mutations that are exclusive to tumour samples. However, I have also found some mutations that are present in controls, but not in any of the tumour samples. Can someone please put some light on the reason behind this?

[jotan]

Are the samples matched? (i.e. control and tumour from a single individual) How are you defining mutation? Excluding known SNPs? Only in coding regions?

[Chris Miller]

To add to these questions - what's the coverage like? If you look at these regions in IGV, are there lots of ambiguously mapped reads that might indicate misalignments?

[Chris Miller]

Also, what kind of somatic mutation caller are you using that reports these sites? Most of the ones I'm aware of clearly label such sites as LOH or Germline mutations, if they report them at all.

Answer 1

[Lemire]

That's probably loss of heterozygosity, not somatic substitutions. Caused by chromosomal regions where one parental copy was loss in the tumor (i.e. deleted). Check if they tend to cluster in regions.

Answer 2

[WouterDeCoster]

Are you sure these positions are references calls or simply unsequenced/absent from the data?