Do strand flips in the dbSNP or training file cause problems for BQSR, Mutect 2, or VQSR?
Our group is working on putting together a file of known germline variants in a non-model organism. While we have a large set of known variants, my colleague has noted that some number of these are reported on the reverse strand of the reference genome rather than the forward strand, such that the alleles have been flipped to the complementary strand.
Will these flipped SNPs cause a problem for BQSR, Mutect 2 variant calling, and VQSR? If these algorithms only need to know the site of the known variation in order to mask it, it seems like we should be fine. If the algorithms actually need to know what the ref and alt alleles are, then it seems like the strand flips would cause a problem.
Thank you for your help in clarifying this!