> The premise is that the amino acid composition is different in a gene
> prior to selection (i.e., before selective pressures arise for the
> protein encoded by the gene) and after selection.
But surely, as soon as a piece of DNA "thinks" of encoding a protein,
selection becomes operative on the protein. Can there really be a period
prior to selection kicking in?
> That's related to my question. You can argue selection cannot be
> present in a gene that isn't functional. But I think it is a
> difficult question to answer, which is why I asked.
What you seem to be missing out is that selection may have acted on
the piece of DNA before it "thinks" of becoming a protein-encoding gene,
(and even after it becomes a protein-encoding gene selection may be
acting on it for genomic reasons discussed elsewhere:
Sincerely, Donald Forsdyke