Exploration of the free-energy surface of a three-helix peptide with Stochastic optimization methods

We report our recent efforts to develop all-atom forcefields that can be used to predict the three-dimensional, tertiary structure of proteins using stochastic optimization methods. We have analyzed an approximate free-energy surface of the 36-residue headpiece of the villin protein and found configurations that were lower in energy than the NMR configuration. We adjusted the parameters of the solvent model to stabilize the NMR structure using a decoy approach. We arrived at a free-energy surface that is characterized by a deep folding funnel populated by different three-helix structures, one of which is very similar to the NMR structure. (C) 2004 Wiley Periodicals, Inc.