Hello all,

I have conducted branch-site analyses to identify positively selected codons on specified branches using PAML.

In contrast to PAML, using branch-specific analyses of codon selection pressures in HYPHY, I regularly find non-signficicant results when applying non-synonymous rate heterogeneity.

Has anyone had similar experiences and, if so, is this logical? I think this finding is unlikely to be widespread if different codon positions are involved in positive selection on an amino acid. Still, many of the codons in the genes I examined are not variable in only one of the three codon positions.

Any ideas on this?

Thanks, Yer

Dear Yer,

I am not sure I understand exactly what is going on with your analyses. PAML finds significant results, and HyPhy doesn't? Also, which HyPhy analysis (and options) did you use to check for lineage and site specific selection? What was the size and divergence level of your alignment?

Generally speaking, fixed effects sites methods in HyPhy tend to be less powerful (for smaller datasets) than the random effects models in PAML, but, because of their non-parametric nature, tend to have lower rates of false positives.

Cheers,
Sergei