Accelerating Systems Biology Computation: Enhanced Sampling of Spatially Realistic Stochastic Models using the Weighted Ensemble Approach

Abstract

We apply the “weighted ensemble” (WE) simulation strategy, previously employed in the context of molecular dynamics and spatially homogeneous chemical kinetics simulations, to a number of three dimensional spatially resolved stochastic systems-biology models. WE is relatively easy to implement, does not require extensive hand-tuning of parameters, does not depend on the details of the underlying simulation algorithm, and can facilitate the unbiased sampling of extremely rare events.We study three systems of varying complexity and structure: a toy model in an artificial geometry with ∼10ˆ3 diffusing molecules, a spatially realistic cellular model with ∼10ˆ6 molecules, and a spatially realistic neuromuscular junction model with time-varying rate “constants” containing ∼10ˆ6 molecules. We are able to speed up the sampling of key events by many orders of magnitude and obtain unbiased estimates of event probabilities. Data is obtained using the publicly available “WESTPA” implementation of WE, which orchestrates MCell kinetic Monte Carlo trajectories of the various models.