ka/ks for comparing more than 2 protein sequences

Question

[lessismore]

Does it make sense to calculate ka/ks comparing more than 2 species (so not comparing just pair of orthologs) to study protein evolution and identify patterns of selection??

If yes, what do you think it is the limit to correctly analyze ka/ks mantaining a reasonable distribution? I've been encountering studies which do this within species with 3+ up to 10 accessions, but I know it is best to do it pairwise so I'm a bit confused on this.

If no, what alternative would you recommend to analyze genetic diversity for a specific gene family in an entire population for that species (or closely related, say >200 individuals) at the gene level.

Thanks for any help

[manaswwm]

AFAIK: The Ka/Ks makes sense in terms of comparing selection on different proteins across phylogenies, I think one of the limitations could be that the more species you consider the poor your alignment could get (more gaps) and hence affect the Ka/Ks. If you compare this to pairwise, then it becomes a little bit like comparing multiple gene alignment with pairwise gene alignment - sure you can get pairwise estimates, but then you could have, for example, certain regions that are not so well-conserved in a subset of species which could influence the overall estimates for entire phylogeny/bunch of closely related species, that you might miss in pairwise alignments. I guess depends a bit also on the question that you are trying o answer.

I have used DataMonkey before for doing such (with multiple sequence alignments) analysis for a couple of proteins in 10-20 species. I do not know if DataMonkey can handle >200 sequences at once, maybe you could also try HyPhy (which is also what DataMonkey implements, but this works locally)