REBIND: Enhancing ground-state molecular conformation via force-based graph rewiring

TL;DR

REBIND introduces a graph rewiring technique based on Lennard-Jones potential to improve deep learning predictions of molecular conformations, especially for atoms influenced by non-bonded interactions, reducing errors by up to 20%.

Contribution

The paper presents a novel graph rewiring method that enhances DL models by explicitly modeling non-bonded atomic interactions in molecular conformation prediction.

Findings

REBIND outperforms existing methods across various molecular sizes.

Achieves up to 20% reduction in prediction error.

Effectively captures non-bonded interactions for low-degree atoms.

Abstract

Predicting the ground-state 3D molecular conformations from 2D molecular graphs is critical in computational chemistry due to its profound impact on molecular properties. Deep learning (DL) approaches have recently emerged as promising alternatives to computationally-heavy classical methods such as density functional theory (DFT). However, we discover that existing DL methods inadequately model inter-atomic forces, particularly for non-bonded atomic pairs, due to their naive usage of bonds and pairwise distances. Consequently, significant prediction errors occur for atoms with low degree (i.e., low coordination numbers) whose conformations are primarily influenced by non-bonded interactions. To address this, we propose REBIND, a novel framework that rewires molecular graphs by adding edges based on the Lennard-Jones potential to capture non-bonded interactions for low-degree atoms. Experimental results demonstrate that REBIND significantly outperforms state-of-the-art methods across various molecular sizes, achieving up to a 20\% reduction in prediction error.