I am generating a consensus sequence for SARS-Cov-2. I am visually inspecting every candidate SNP in a pileup. Sometimes I find that there are two consistent variants for the same position, but I can only generate one consensus sequence. What is the best practice? Getting always the majoritarian variant or getting the minoritarian one if I find that this SNP is under-reported in other public samples? apart of which one is best practice, are both practices legit?
Making a consensus sequence tends to involve some arbitrary decisions. In this case, the common practices are:
- Use the ancestral allele (this isn't always apparent)
- Use the most common allele
- Randomly choose
I think 1 and 2 are generally preferred, with 2 being easier to implement.