All,
Without going into too much detail, and apologising for not having read
everybody's contribution to this discussion too carefully, I pass some comments
on what has been said.
Firstly, it strikes me that the secondary structure prediction as performed
by Srinivasan and Rose *is* really very good. Whether this will stand up to
closer scrutiny remains to be seen. If it does, then I think it says
something about the prediction of secondary structure by way of predicting
supersecondary structure. Their approach may well reflect the need to look
beyond the immediate environment of a secondary structure element (i.e. by
a careful consideration of the adjacent sequence), which everyone has
admitted for such a long time.
Secondly, it would seem that they have slightly oversold this method. It does
not seem to be to be a method for predicting whole tertiary structures. Rather
appears to be a very promising method for the prediction of supersecondary
structure, or of small folding motifs. It seems very impressive to me that
such a simple method, using (seemingly) so little empirical data can predict
local chain conformation even sometimes. They could do worse than to pursue
this possibility further.
Rob
Robert B. Russell Biomolecular Modelling, Imperial Cancer Research Fund
44 Lincoln's Inn Fields, P.O. Box 123, London, WC2A 3PX, U.K.
Tel: 44 171 269 3583 FAX: 44 171 269 3479 e-mail: russell at icrf.icnet.uk
WWW http://bonsai.lif.icnet.uk
