Tolerances for non cells-of-interest for somatic variant calling

[rjohnp]

Hi,

we are looking at several types on non-solid tumors, so it is difficult to get a 'pure' sample of cancer cells for exome sequencing.

I was wondering what sort of level of purity of cancer cells for subsequent DNA we will need in order to identify acquired variants in our cancer cells of interest. I know it will vary with calling pipeline and variable tolerances for the software, but I'm just looking for a ballpark idea of how pure it needs to be for decent sensitivity without a flood of false positives.

Thanks for any help