Small Protein Folding using Weighted Ensemble Simulation: A Case Study of Trp-Cage

Abstract

The 20-residue protein Trp-cage is a rapidly folding protein ideal for testing different strategies in folding simulations. The system was studied via the weighted ensemble (WE) approach where trajectories evolved according to Langevin dynamics in implicit solvent. In the WE method [Huber and Kim, Biophys. J., 1996], the configuration space is divided into arbitrary regions called bins. Trajectories arriving to new bins are replicated into identical daughter trajectories that inherit the history and a corresponding fraction of the parent's weight. In general, WE is a parallel method that forces a system to explore timescales and regions of configuration space typically inaccessible in regular simulations. Unbiased estimates of rates and other properties are produced. Appropriate selection of bins remains a challenging aspect of WE, however, especially in non-trivial systems such as proteins. For Trp-cage, different binning strategies are explored using 1D and 2D binning coordinates with the goal of improving efficiency. Rate constants calculated in both the high and low friction regimes are compared with independent estimates.