This site is now read-only. You can find our new documentation site and support forum for posting questions here.
Be sure to read our welcome blog!
How to set GVCF genotypes too ./. based on the GQ score
I have a reasonably large non-human multi-VCF dataset containing ~280 samples and ~70M variants. I want to filter low quality genotype calls (but not variants as a whole). This does not seem to be standard procedure - most filtering seems to operate at the variant level and I don't understand why that is.
I would like to take eg all genotypes where the GQ is <= 5 and set them to ./.. Is that possible, or should I write my own code to do that? Are there any caveats with using GQ? My understanding is that the PL calculation weighs all the evidence in predicting likelihoods, and GQ as the delta between the best and next best genotype, is the best indicator of whether the best genotype is likely to be correct,