Predictive Collective Variable Discovery with Deep Bayesian Models

TL;DR

This paper introduces a Bayesian inference framework for discovering collective variables in atomistic systems, enabling better sampling and understanding of complex biochemical and materials science processes.

Contribution

It formulates CV discovery as a Bayesian inference problem, leveraging machine learning and variational inference to identify physically meaningful CVs with confidence estimates.

Findings

Successfully applied to alanine dipeptide and peptide systems

Generated CVs related to key physicochemical properties

Provided probabilistic confidence in the CVs' predictive ability

Abstract

Extending spatio-temporal scale limitations of models for complex atomistic systems considered in biochemistry and materials science necessitates the development of enhanced sampling methods. The potential acceleration in exploring the configurational space by enhanced sampling methods depends on the choice of collective variables (CVs). In this work, we formulate the discovery of CVs as a Bayesian inference problem and consider the CVs as hidden generators of the full-atomistic trajectory. The ability to generate samples of the fine-scale atomistic configurations using limited training data allows us to compute estimates of observables as well as our probabilistic confidence on them. The methodology is based on emerging methodological advances in machine learning and variational inference. The discovered CVs are related to physicochemical properties which are essential for understanding mechanisms especially in unexplored complex systems. We provide a quantitative assessment of the CVs in terms of their predictive ability for alanine dipeptide (ALA-2) and ALA-15 peptide.