Folding Proteins from First-Principles.
Feb 6, Mon, 2006
3:30pm - 4:30pm, 1001 Malott
The protein folding problem is of fundamental importance in molecular biology. Knowledge of how and why proteins fold is critical in understanding function, protein design and misfolding. Detailed computer simulations from first-principles can provide important insights in this understanding. However, conventional brute-force simulations, particularly those with explicit solvent, are severely limited by both the sampling capacity and force field accuracy. In this study, we address both obstacles by carefully re-parameterizing an efficient generalized Born (GB) implicit solvent force field. The optimization is guided by pair-wise interactions between polar groups in explicit solvent and by extensive folding and unfolding simulations of small peptides. An advanced sampling technique, named the replica exchange method (REX), is used to speed up the convergence of thermodynamics properties. The resulting GB force field correctly predicts the experimental conformational equilibria of both helical peptides and beta-hairpins. It is also the first single consistent force field to fold both designed sequences trpzip2 and Trp-cage. We further demonstrate that important lessons on the folding mechanisms of beta-hairpins can be learned from these simulations.