Protein structure prediction by global optimization of a potential energy function

An approach based exclusively on finding the global minimum of an appropria te potential energy function has been used to predict the unknown structure s of five globular proteins with sizes rang;ng from 89 to 140 amino acid re sidues. Comparison of the computed lowest-energy structures of two of them (HDEA and MarA) with the crystal structures, released by the Protein Data B ank after the predictions were made, shows that large fragments (61 residue s) of both proteins were predicted with rms deviations of 4.2 and 6.0 Angst rom for the Ca atoms, for HDEA and MarA, respectively. This represents 80% and 53% of the observed structures of HDEA and MarA respectively. Similar r ms deviations were obtained for similar to 60-residue fragments of the othe r three proteins. These results constitute an important step toward the pre diction of protein structure based solely on global optimization of a poten tial energy function for a given amino acid sequence.