Milestoning network refinement by incorporating experimental thermodynamic and kinetic data

TL;DR

This paper introduces a refinement method for Milestoning kinetic networks that integrates experimental thermodynamic and kinetic data to improve the accuracy of rare event simulations.

Contribution

It presents a novel approach to refine Milestoning networks by minimizing divergence while incorporating experimental data, enhancing quantitative agreement with experiments.

Findings

Refined kinetic networks better match experimental data.

Method applied successfully to ligand binding/unbinding in β-cyclodextrin.

Improved accuracy in rare event kinetics calculations.

Abstract

Milestoning is an accurate and efficient method for rare event kinetics calculations by constructing a continuous-time kinetic network connecting the reactant and product states. However, even with adequate sampling, its accuracy can also be limited by the force fields, which makes it challenging to achieve quantitative agreement with experimental data. To address this issue, we present a refinement approach by minimizing the Kullback-Leibler divergence rate between two Milestoning networks while incorporating experimental thermodynamic (equilibrium constants) and kinetic (rate constants) data as constraints. This approach ensures that the refined kinetic network is minimally perturbed with respect to the original one, while simultaneously satisfying the experimental constraints. The refinement approach is demonstrated using the binding and unbinding dynamics of a series of six small molecule ligands for the model host system, $\beta$-cyclodextrin.