Simulations of molecular dynamics with classical mechanics and simple empirical force fields give the most accurate representation outside of quantum mechanics, and can today be performed on systems containing many thousands of atoms for periods of simulated time of many nanoseconds. Simulation accuracy has been greatly improved by economical methods for computing long-range electrostatic forces (by our colleagues Board at Duke and Darden at NIEHS). Computer resources have grown so that it is feasible to implement so-called steered molecular dynamics, SMD (implemented at UNC with use of vmd, a molecular graphics program from Klaus Schulten's group at Illinois). In our SMD project we are studying the extraction of a non-covalently bound small molecule from a protein, by a user-specified tug, through any of several likely exit routes, and this is followed by batch calculations of the energetics experienced by a molecule moving naturally along such a path, i.e., by diffusion, rather than as the result of an artificial tug.

Changes in conformation and interactions with small molecules are, of course, intrinsic to the central role played by proteins in biology. We have over the years developed methods to study the energetics of these processes. This work has, for example, shown that the simulations reproduce the experimentally measured equilibria between helix and coil states of polypeptides, and give insight into the interactions that can be held responsible for stabilizing the helix. Rather surprisingly, our results indicate that electrostatic forces contribute nothing to the stability of the helix relative to the coil, but that it is the balance of the van der Waals (packing) forces which favors the helix over the coil. Recently, we have been able to reproduce the binding equilibrium of a small molecule to a protein in terms of its free energy, and to decompose this quantitatively into parts, a binding energy and a "cratic" free energy, the latter corresponding to the loss of positional and orientational freedom when two molecules form a complex.

Back to Workshop Schedule

1996-1997 Mathematics in High Performance Computing