Stochastic Optimization Methods for Protein Folding ALEXANDER SCHUG, ABHINAV VERMA, KYU HWAN LEE, WOLFGANG WENZEL Forschungszentrum Karlsruhe, Institut für Nanotechnologie, P.O. Box 3640, 76021 Karlsruhe, GERMANY Forschungszentrum Karlsruhe, Institut für Nanotechnologie, P.O. Box 3640, 76021 Karlsruhe, GERMANY Supercomputing Materials Laboratory Korean Institute for Science and Technology Seoul KOREA http://www.fzk.de/biostruct Abstract: - We recently developed an all-atom free energy forcefield (PFF01) for protein structure prediction with stochastic optimization methods. We demonstrated that PFF01 correctly predicts the native conformation of several proteins as the global optimum of the free energy surface. Here we review recent folding studies, which permitted the reproducible all-atom folding of the 20 amino-acid trp-cage protein, the 40-amino acid three-helix HIV accessory protein and the sixty amino acid bacterial ribosomal protein L20 with a variety of stochastic optimization methods. These results demonstrate that all-atom protein folding can be achieved with present day computational resources for proteins of moderate size. Keywords: protein folding, stochastic optimization 1. Introduction Ab-initio protein tertiary structure prediction (PSP) and the elucidation of the mechanism of the folding process are among the most important outstanding problems of biophysical chemistry [1,2]. The many complementary proposals for PSP span a wide range of representations of the protein conformation, ranging from coarse grained models to atomic resolution. The choice of representation often correlates with the methodology employed in structure prediction, ranging from empirical potentials for coarse grained models [3,4] to complex atom-based potentials that directly approximate the physical interactions in the system. The latter offer insights into the mechanism of protein structure formation and promise better transferability, but their use incurs large computational costs that has confined all-atom protein structure prediction to all but the smallest peptides [5,6]. It has been one of the central paradigms of protein folding that proteins in their native conformation are in thermodynamic equilibrium with their environment [7]. Exploiting this characteristic the structure of the protein can be predicted by locating the global minimum of its free energy surface without recourse to the folding dynamics, a process which is potentially much more efficient than the direct simulation of the folding process. PSP based on global optimization of the free energy may offer a viable alternative approach, provided that suitable parameterization of the free energy of the protein in its environment exists and that global optimum of this free energy surface can be found with sufficient accuracy [8]. We have recently demonstrated a feasible strategy for all-atom protein structure prediction [9,10,11] in a minimal thermodynamic approach. We developed an all- atom free-energy forcefield for proteins (PFF01), which is primarily based on physical interactions with important empirical, though sequence independent, corrections [11]. We already demonstrated the reproducible and predictive folding of four proteins, the 20 amino acid trp-cage protein (1L2Y) [9,12], the structurally conserved headpiece of the 40 amino acid HIV accessory protein (1F4I) [10,13] and the sixty amino acid bacterial ribosomal protein L20 [14]. In addition we showed that PFF01 stabilizes the native conformations of other proteins, e.g. the 52 amino-acid Proceedings of the 5th WSEAS Int. Conf. on SIMULATION, MODELING AND OPTIMIZATION, Corfu, Greece, August 17-19, 2005 (pp477-481)