I have a list of variants that I would like to check exist in some cell-free DNA extracted from plasma. These variants arise from DNA extracted from clinical tissue, and tumour-derived variants may appear at very low levels in the cell-free DNA extracted from the same patients. The sequencing is such (using UMIs and consensus deduplication) that one or two variant-containing fragments post-deduplication would potentially be enough to confirm the presence of ctDNA in the plasma.
I have tried running the GATK pipeline and the samtools mpileup/mutscan pipeline, and both are excellent in extracting variants with allelic fractions close to 50%, but these are likely derived from genomic cfDNA rather than circulating tumour DNA.
Is there a bioinformatic method sensitive enough to detect the one or two DNA molecules containing the variant?
If you just need to call low frequency variants, why does it have to be directly from FASTQ files?
I was looking for a method sensitive enough for a single read that covers the variant
There are many variant callers that allow you to set thresholds (one nice overview). If you just want to check for any mismatches, you can use pileups directly. It's not necessary to add the extra caveats to deal with FASTQs directly.