Effective Data-Driven Collective Variables for Free Energy Calculations from Metadynamics of Paths

TL;DR

This paper introduces a novel method to generate high-quality datasets for machine learning-based collective variables using metadynamics of paths, improving free energy landscape predictions in molecular simulations.

Contribution

It demonstrates how to generate datasets via path metadynamics to create effective ML-based CVs, enhancing free energy calculations for complex molecular processes.

Findings

Successful application to a 2D model potential

Effective prediction of alanine dipeptide isomerization

Improved accuracy of free energy landscape estimation

Abstract

A variety of enhanced sampling methods predict multidimensional free energy landscapes associated with biological and other molecular processes as a function of a few selected collective variables (CVs). The accuracy of these methods is crucially dependent on the ability of the chosen CVs to capture the relevant slow degrees of freedom of the system. For complex processes, finding such CVs is the real challenge. Machine learning (ML) CVs offer, in principle, a solution to handle this problem. However, these methods rely on the availability of high-quality datasets -- ideally incorporating information about physical pathways and transition states -- which are difficult to access, therefore greatly limiting their domain of application. Here, we demonstrate how these datasets can be generated by means of enhanced sampling simulations in trajectory space via the metadynamics of paths [arXiv:2002.09281] algorithm. The approach is expected to provide a general and efficient way to generate efficient ML-based CVs for the fast prediction of free energy landscapes in enhanced sampling simulations. We demonstrate our approach with two numerical examples, a two-dimensional model potential and the isomerization of alanine dipeptide, using deep targeted discriminant analysis as our ML-based CV of choice.