Thorsten Burmester <burmeste at mail.Uni-Mainz.de> writes:
> Does anybody know why the synonymous and non-synonymous substition rates
> are correlated in many genes? I.e., in highly conserved genes one may
> find that both Ka and Ks are strongly reduced when compared to a less
> conserved paralogous gene.
Comeron and Kreitman's "The correlation between synonymous and
nonsynonymous substitutions in Drosophila: Mutation, Selection, or
Relaxed Constraints" (Genetics 1998) would be a great paper for you.
One suggestion (Akashi 1994) is that the correlation is a result of
codon-level selection for translational accuracy. Any amino acid
position that has high functional importance (and hence low Ka) will
also need to be translated accurately, thus leading to selection at
the codon level (and hence low Ks).
Another suggestion (Wolfe and Sharp 1993) is that the pattern results
from "doublet" mutations. The idea is that these mutations would
commonly change the third position of one codon (causing a syn change)
and the first of an adjacent codon (causing a nonsyn change), thereby
leading to a correlation between Ka/Ks.
Another suggestion (Lipman and Wilbur 1985) is that after a
nonsynonymous substituion, synonymous substitution for the preferred
codon will follow, thus generating the observed correlation.
In Comeron and Kreitman, they present evidence that rejects all three.
However, the first suggestion may still work for amino acid sites with
Department of Integrative Biology
University of California
Berkeley, CA 94720-3140