I think that the concept of "transmembrane usage ratio" is biased. The reason:
Some transmembrane protein domains specifically serve the purpose to place
select polar amino acid sidechains into this environment, for example in
transmembrane channels.
Witness the values for the Glu/Gln and Asp/Asn pairs which make the charged
sidechains appear less hydrophobic than their neutral counterparts. I would
argue that those are in transmembrane segments for a very specific function and
thus distort a classification that should reflect physico-chemical properties
only.
I find the concept interesting, though, if only to provoke some thought...
Christoph
-- | Christoph Weber Sen. Research Associate | Dept.of Molecular Biology, MB2 619-554-7283 or -8754 (phone) | The Scripps Research Institute 619-554-3757 (FAX) | La Jolla CA 92037 weber@scripps.edu | http://www.scripps.edu/~chazin/people/cw.html