EquiBind: Geometric Deep Learning for Drug Binding Structure Prediction

TL;DR

EquiBind is a geometric deep learning model that predicts drug-protein binding structures directly, offering faster and more accurate results than traditional methods, with a novel fine-tuning approach for ligand conformations.

Contribution

The paper introduces EquiBind, a SE(3)-equivariant deep learning model for direct prediction of binding sites and ligand poses, reducing computational costs and improving accuracy.

Findings

EquiBind outperforms traditional docking methods in speed and accuracy.

Coupling EquiBind with fine-tuning techniques further improves results.

A new fast fine-tuning model based on closed-form solutions enhances ligand conformation adjustment.

Abstract

Predicting how a drug-like molecule binds to a specific protein target is a core problem in drug discovery. An extremely fast computational binding method would enable key applications such as fast virtual screening or drug engineering. Existing methods are computationally expensive as they rely on heavy candidate sampling coupled with scoring, ranking, and fine-tuning steps. We challenge this paradigm with EquiBind, an SE(3)-equivariant geometric deep learning model performing direct-shot prediction of both i) the receptor binding location (blind docking) and ii) the ligand's bound pose and orientation. EquiBind achieves significant speed-ups and better quality compared to traditional and recent baselines. Further, we show extra improvements when coupling it with existing fine-tuning techniques at the cost of increased running time. Finally, we propose a novel and fast fine-tuning model that adjusts torsion angles of a ligand's rotatable bonds based on closed-form global minima of the von Mises angular distance to a given input atomic point cloud, avoiding previous expensive differential evolution strategies for energy minimization.