Extensive Evaluation of Weighted Ensemble Strategies for Calculating Rate Constants and Binding Affinities of Molecular Association/Dissociation Processes

Abstract

The weighted ensemble (WE) path sampling strategy is highly efficient in generating pathways and rate constants for rare events using atomistic molecular dynamics simulations. Here we extensively evaluated the impact of several advances to the WE strategy on the efficiency of computing association and dissociation rate constants (kon, koff) as well as binding affinities (KD) for a set of benchmark systems, listed in order of increasing timescales of molecular association/dissociation processes: methane/methane, Na+/Cl-, and K+/18-crown-6 ether. In particular, we assessed the advantages of carrying out (i) a large set of \"light-weight\" WE simulations that each consist of a small number of trajectories vs. a single \"heavy-weight\" WE simulation that consists of a relatively large number of trajectories, (ii) equilibrium vs. steady-state WE simulations, (iii) history augmented Markov State Model (haMSM) post-simulation analysis of equilibrium sets of trajectories, and (iv) tracking of trajectory history (the state last visited) during the dynamics propagation of equilibrium WE simulations. Provided that state definitions are known in advance, our results reveal that heavy-weight, steady-state WE simulations are the most efficient protocol for calculating kon, koff, and KD values. If states are not strictly defined in advance, heavy-weight, equilibrium WE simulations are the most efficient protocol. This efficiency can be further improved with the inclusion of trajectory history during dynamics propagation. In addition, applying the haMSM post-simulation analysis enhances the efficiency of both steady-state and equilibrium WE simulations. Recommendations of appropriate WE protocols are made according to the goals of the simulations (e.g. to efficiently calculate rate constants and/or generate a diverse set of pathways).