Filtering for pyrho

[selasphoruskershaw]

Hello,

Are there any best-practice filtering strategies of VCFs for use in calculating recombination rate in pyrho? For example, setting a minor allele frequency threshold, the use of indels, etc?

Thank you!

[jeffspence]

Hello,

I have not done much benchmarking on how filtering of VCFs affects estimated recombination rates. pyrho automatically ignores everything except bi-allelic SNPs, so no need to filter indels. For the paper, I removed SNPs that were in regions that are difficult to map to and SNPs with more than a few missing genotypes, but otherwise did not filter on MAF or anything else. In a well-QC'd publicly available human dataset that seemed to work pretty well, but of course messier data might require more strict filtering.

Inference is quite fast using pyrho (as long as there is not too much missing data) so it might be good to try out a few filters and see if the results are concordant.

Hope this helps!