| Abstract: We introduce a new method for constraint-based sampling of protein configurations, related to the previously reported FRODA method [1]. The task is to sample configurations of the atoms that satisfy a set of geometric constraints. The key component of the new approach is an explicit constraint-enforcing energy function that is zero if the constraints are met but increases quadratically as constraints are violated. Acceptable configurations therefore lie on the flat valley floor of the energy landscape. We sample the configurations that meet the constraints by perturbation from an initial acceptable configuration, followed by conjugate-gradient minimization back down to the valley floor, arriving at a new acceptable configuration. A momentum-like perturbation strategy facilitates exploration of large collective motions. Compared to the FRODA method, the new approach is able to restore constraints quicker and more robustly, and to tighter tolerances. We also discuss current benchmarking efforts to compare the rate of exploration of phase space using the constraint-based sampling method against that attained using molecular dynamics.
1. Wells, S., Menor, S., Hespenheide, B.M., and Thorpe, M. F. Constrained geometric simulation of the diffusive motions in proteins. Phys Bio 2005;2:S127-S136 |
