Molecular Free Energies, Rates, and Mechanisms from Data-Efficient Path Sampling Simulations

Abstract

Molecular dynamics is a powerful tool for studying the thermodynamics and kinetics of complex molecular events. However, these simulations can rarely sample the required time scales in practice. Transition path sampling overcomes this limitation by collecting unbiased trajectories and capturing the relevant events. Moreover, the integration of machine learning can boost the sampling while simultaneously learning a quantitative representation of the mechanism. Still, the resulting trajectories are by construction non-Boltzmann-distributed, preventing the calculation of free energies and rates. We developed an algorithm to approximate the equilibrium path ensemble from machine-learning-guided path sampling data. At the same time, our algorithm provides efficient sampling, mechanism, free energy, and rates of rare molecular events at a very moderate computational cost. We tested the method on the folding of the mini-protein chignolin. Our algorithm is straightforward and data-efficient, opening the door to applications in many challenging molecular systems.

Collective Variable-Based Enhanced Sampling: From Human Learning to Machine Learning

“…However, emerging studies suggest that the potential of machine learning is considerable. Recent studies have begun to combine path-sampling algorithms with machine-learning techniques to leverage the strengths of both, and the performance of these hybrid methods has been remarkable. ,,− Nevertheless, despite evidence supporting the potential of machine-learning algorithms, the empirical setting of critical parameters (e.g., lag time τ), which significantly impact the quality of the resultant mCVs, makes training mCVs akin to unboxing a blind box. Further exploration into rational hyperparameter selection, possibly via meta-learning strategies that enable algorithms to “learn to learn”, is crucial.…”

Data-Driven Path Collective Variables

“…This was later expanded to obtain CVs with maximal transmission coefficients or using cross-entropy minimization and regularization . Recently, this approach has been recast in a reinforcement learning framework to allow the identification of CVs over the course of transition path sampling simulations, , including an extension to extract free energies and rates . Committor-based methods have also been combined with slow process identification …”

Sampling a Rare Protein Transition Using Quantum Annealing

“…Low acceptance rates often limit the simulation of complex transitions. To tame this issue, a promising integration with machine learning has been introduced to optimize this procedure, , enabling the simulation of rare and complex processes, such as membrane-protein assembly. However, TPS still requires sampling the duration of the transition with unbiased simulations, which can be very expensive.…”