Simulations meet Machine Learning in Structural Biology

TL;DR

This paper discusses how integrating machine learning with classical and quantum simulations can significantly enhance prediction accuracy and efficiency in structural biology and drug discovery.

Contribution

It proposes leveraging ML models trained on simulation data to overcome current limitations in simulation speed and accuracy in structural biology.

Findings

ML can improve prediction accuracy in structural biology

ML reduces time-to-prediction compared to classical methods

Synergies between simulations and ML can transform drug discovery

Abstract

Classical molecular dynamics (MD) simulations will be able to reach sampling in the second timescale within five years, producing petabytes of simulation data at current force field accuracy. Notwithstanding this, MD will still be in the regime of low-throughput, high-latency predictions with average accuracy. We envisage that machine learning (ML) will be able to solve both the accuracy and time-to-prediction problem by learning predictive models using expensive simulation data. The synergies between classical, quantum simulations and ML methods, such as artificial neural networks, have the potential to drastically reshape the way we make predictions in computational structural biology and drug discovery.