Choice of adaptive sampling strategy impacts state discovery, transition probabilities, and the apparent mechanism of conformational changes

• DOI: 10.48550/arxiv.1805.04616
• Publication Date: 2018-01-01
• Authors: Maxwell I. Zimmerman; Justin R. Porter; Xianqiang Sun; Roseane R. Silva; Gregory R. Bowman
• Keywords: Traverse; Adaptive sampling; Rare events; Kinetic Monte Carlo

Interest in equilibrium-based sampling methods has grown with recent advances computational hardware and Markov state modeling (MSM) methods, yet outstanding questions remain that hinder widespread adoption. Namely, how do strategies explore conformational space might this influence predictions? Here, we seek to answer these for four commonly used methods: 1) a long simulation, 2) many short simulations, 3) adaptive sampling, 4) FAST. We first develop theoretical framework analytically calculating the probability of discovering states uncover drastic effects varying number length simulations. then use kinetic Monte Carlo simulations on variety physically inspired landscapes characterize discovery transition pathways. Consistently, find FAST discover target highest traverse realistic Furthermore, pathology parallel sometimes predict an incorrect pathway by crossing large energy barriers would typically circumnavigate, which refer as tunneling. To protect against tunneling, introduce FAST-string, samples along highest-flux paths refine MSMs probabilities discriminate between competing Additionally, compare MSM estimators describing thermodynamics kinetics. For recommend normalizing counts out each after adding pseudo-counts avoid creating sources or sinks. Lastly, evaluate our insights from simple all-atom molecular dynamics folding {\lambda}-repressor protein. FAST-contacts predicts same but orders magnitude less simulation time.