Hey folks,

I'm a new user with a complicated question.

I'm working on sequence evolution in Drosophila, a system in which a lot of attention has been paid to codon bias and the fact that, in Drosophila evolution, not all synonymous substitutions are created equal. Codon bias per se doesn't seem to be too much of an issue in that we are usually estimating the expected codon frequencies from the data. However, codon biases stronger in some Drosophila species than others, and there is evidence that codon usage preferences can change over evolutionary time. This suggests to me that a "truer" model would be one in which base/codon frequencies are estimated independently along lineages and/or an additional (lineage-specific) parameter be added to the transition matrix to highlight the difference between switches from preferred to unpreferred, synonymous substitutions.

My questions for y'all:
1) Are these sensible solutions?
2) How might I go about modifying a script like YangNielsenBranchSite.bf (or some non-branch specific model) to accommodate this issue?

Thanks,

David Garfield
European Molecular Biology Laboratory

PS: Eventually I'm hoping to incorporate this all into a Wong and Nielsen/Fedrigo and Haygood type analysis on cis-regulatory evolution, so the correct answer here might be "don't freak out over your coding models".