(1) Department of Physical Chemistry, Department of Biological Chemistry, the
Fritz Haber Research Center for Molecular Dynamics and the Wolfson Center
for Applied Structural Biology, Hebrew University, Givat Ram, Jerusalem
91904, ISRAEL.
(2) Department of Molecular Biology, Scripps Reasarch Institute, La Jolla, CA
92037, USA
Homology-based modeling and global optimization of energy functions are two complementary approaches to predict protein structures. A combination of the two approaches is proposed. It is shown that coupling the search for the native folds (assumed to be the global energy minimum) of several homologous proteins, is significantly more successful than the folding of a single protein. The reason is the elimination of local minima not shared by all homologous proteins and better sampling of the relevant conformation space. Structure prediction of two protein families, pancreatic peptides and homeodomains, is used to test the proposed approach.