I have cfDNA samples which were sequenced (at about 300-1000X) on a 30-gene panel via a capture-seq protocol for the purposes of calling variants on circulating tumour DNA.
However, the resulting variant calls (from both varscan and Mutect2) are dominated by transversions from C>A and G>T at low allele frequencies, which are characteristic of DNA oxidation of guanine to 8-oxoguanine, and which I suspect are artefacts.
I have matched baseline tumour samples from the same patients, sequenced in the same way, (and there are a few C>A/G>T transversions at low AFs in these samples that I can filter out by a direct allele fraction filter).
My question is this, what is the best way of filtering out the false positive mutations derived from the 8-oxoG formation and retaining the true positives (which will be at low allelic fraction)?
Some background: I do have serial (cfDNA) time points for some patients, so I am loath to just look at mutations called in the baseline tumour samples, as any (true) novel variants appearing later in the time points would be missed.