Dear Sergei,

Hopefully you can find the time to answer this question about the selection tests applied to population samples.

Phylogeny-type tests of selection pressures typically are affected by differences in codon usage across taxa (i.e., they do not handle these difference well if they are substantial), how do you think this affects the population comparison of selection pressures across sites?

If different dN/dS patterns are present throughout the population trees and codons are used with different frequencies in the populations, then this should not seriously affect the test given that the dS and dN rates are estimated for each populations separately (and if other processes such as mutation pressures towards GC- or AT-rich genomes are neglected).

But even if mutation processes differ between populations, as long as different amino acids are selected for in the two populations (another possible consequence of codon usage bias differences), the test does not produce false positive, because it is the same type of selection pressure.

In all, I think the population test should be much more robust to differences in codon bias usage than phylogeny type of tests. Do you agree?

Thanks again, Yer