Efficient Simulation of Complex Transition Pathways using a Weighted Ensemble Based String Method

Abstract

Large-scale conformational transitions in biomolecules occur on complex, high-dimensional free energy surfaces. Simulating these transitions is often challenging due to the mismatch between the timescales required to traverse the energetic landscape and those accessible by conventional simulation techniques. The “weighted ensemble” (WE) path sampling method is a rigorous technique for simulating this class of rare transitions. We extend the WE method by combining it with a string method to adaptively refine the set of order parameters used to enhance sampling along the transition pathway. This permits sampling transitions in the space of many order parameters for a wide range of equilibrium and non-equilibrium processes. From these simulations, accurate estimates of steady state conformational distributions and reaction rates can be obtained, even for systems with complex transition pathways that may involve metastable intermediates. We demonstrate the application of this method using two simple models of driven Brownian motion, protein conformational change with a coarse-grained model, and fully-atomistic models.