search

popgenmethods / pyrho

Fast inference of fine-scale recombination rates based on fused-LASSO
MIT License
44 stars 6 forks source link

multiple chromosomes, unique SFS or WG? #19

Closed cistarsa closed 3 years ago

cistarsa commented 3 years ago

Hello, and thank you for this super useful and efficient program. I'm generating recombination maps for my system's whole genome (scaffold n=190) and was wondering if I can use a demographic reconstruction from the entire genome or should I generate SFS for each chromosome, then use the subsequent reconstruction for table generation? Any suggestions would be hugely helpful, thank you.

jeffspence commented 3 years ago

It's up to you, but my inclination would be to use the genome-wide SFS. If you build the SFS genome-wide you'll have more data so the statistical noise will be smaller. It's true that there could be differences in e.g., background selection, which might look like different per-chromosome demographic histories, but I would hope that those differences would be small. Also, pyrho assumes a neutral model, so if selection is pervasive enough to dramatically change inferred demographic histories across different chromosomes, then I would definitely be careful in interpreting the results regardless of whether per-chromosome demographies are inferred.

An important exception would be something like sex chromosomes where you really expect the number of copies of that chromosome in the population to behave in a chromosome-specific way.

Hope this helps!

cistarsa commented 3 years ago

Great, that's what I was assuming as well. Very helpful response thank you!