Hello! Newbie to bioinformatics, here. I'm comparing the GATK-For-Microbes workflow and BCFTools on bacterial WGS data. Strangely, both mostly returned different variants after filtering. I recognize GATK isn't the preferred variation caller on bacterial data, but it's widely used for variation calling on microbial data – so I find the difference in results interesting.
Anyways.... a colleague has asked me to filter out the results based on the allele fraction of each site. This has me wondering: how important is the allele fraction for a site when determining if a variant exists or not? I've noticed that, post-filtering, many of the variants have an allele fraction less than ~.5. Experimentally, why wouldn't an allele be present on most of the reads for a site that we deem a true variant?
Thank you!
The allele fraction is the main thing but only after artefactual alignments have been removed. If the AF is .2 but the other .8 of reads should have been aligned somewhere else, then it's likely a true variant. That cuts both ways, of course.