Potential role of long molecular dynamics simulations in exploring protein folding dynamics


David E. Shaw, D. E. Shaw Research and Columbia University, 120 W. 45th St., 39th Floor, New York, NY 10036
Our research group is currently building a specialized, massively parallel machine called Anton which, when completed late this year, should be capable of executing millisecond-scale molecular dynamics simulations of proteins -- about three orders of magnitude beyond the duration of the longest current MD simulations. This talk will consider the question of how such simulations might contribute to the understanding of protein folding dynamics. Although this question cannot yet be fully answered, several possible folding-related applications will be examined, with a particular focus on application-specific factors affecting the relative merits of a few very long MD runs vs. a very large number of much shorter runs. Also discussed will be the very real risk that serious deficiencies may be discovered in the functional accuracy of millisecond-scale simulated trajectories that have not been observed in nanosecond- to microsecond-scale simulations, along with the potential implications of such an outcome.