Searching for a specific somatic mutation in Illumina data

Question

[manfi]

I am analysing some DNA samples extracted from human tissue with a disease of interest and from non-diseased controls. My hypothesis is that the diseased samples might contain specific somatic mutations. I have already analysed the samples using ddPCR and the results suggest that one mutation of interest might be present with a very low frequency (0.2%). While ddPCR has a certain degree of artifacts, there's a statistically significant difference compared to the non-diseased samples.

I have also got Illumina targeted sequencing data (Sureselect, with UMIs, coverage in the range 300x - 10000x), where so far, I haven't been able to detect the mutations of interest using Mutect2 in standard settings (however, I understand that 0.2% is difficult to detect). Is there a good method to look for specific mutations? My question would be something like this: how frequent is A->G in base 2500? I'd like to obtain this information from about 40 samples and see whether the results correlate with the ddPCR results.

Thanks!

Answer 1

[Shred]

You could specify a target region in Mutect2 using the -L argument, to speed up the analysis.

I'd suggest to use --bam-output to inspect assembled haplotypes manually from reads using a bunch of samples, to verify the presence of the mutation. This could guide you on choosing the best performing parameters to apply on all the cohorts. Remind that mutect2 does downsampling when provided with very high coverage (like in your case): this could definitively impact your analysis. Try to set --max-reads-per-alignment-start to 0 to consider all the reads.