Learning Gradient Fields for Molecular Conformation Generation

Abstract

We study a fundamental problem in computational chemistry known as molecular conformation generation, trying to predict stable 3D structures from 2D molecular graphs. Existing machine learning approaches usually first predict distances between atoms and then generate a 3D structure satisfying the distances, where noise in predicted distances may induce extra errors during 3D coordinate generation. Inspired by the traditional force field methods for molecular dynamics simulation, in this paper, we propose a novel approach called ConfGF by directly estimating the gradient fields of the log density of atomic coordinates. The estimated gradient fields allow directly generating stable conformations via Langevin dynamics. However, the problem is very challenging as the gradient fields are roto-translation equivariant. We notice that estimating the gradient fields of atomic coordinates can be translated to estimating the gradient fields of interatomic distances, and hence develop a novel algorithm based on recent score-based generative models to effectively estimate these gradients. Experimental results across multiple tasks show that ConfGF outperforms previous state-of-the-art baselines by a significant margin.

arXiv commentICML 2021, Long talk

References in corpus (14)

• Semi-Supervised Classification with Graph Convolutional Networks
• Neural Message Passing for Quantum Chemistry
• Deep Potential Molecular Dynamics: a scalable model with the accuracy of quantum mechanics
• Molecular Graph Convolutions: Moving Beyond Fingerprints
• Quantum-Chemical Insights from Deep Tensor Neural Networks
• Junction Tree Variational Autoencoder for Molecular Graph Generation
• Tensor field networks: Rotation- and translation-equivariant neural networks for 3D point clouds
• sGDML: Constructing Accurate and Data Efficient Molecular Force Fields Using Machine Learning
• Strategies for Pre-training Graph Neural Networks
• E(n) Equivariant Graph Neural Networks
• Relevance of Rotationally Equivariant Convolutions for Predicting Molecular Properties
• Learning Neural Generative Dynamics for Molecular Conformation Generation
• MARS: Markov Molecular Sampling for Multi-objective Drug Discovery
• TorsionNet: A Reinforcement Learning Approach to Sequential Conformer Search

Cited by in corpus (2)

• A Comprehensive Survey on Deep Graph Representation Learning
• Time Series Diffusion Method: A Denoising Diffusion Probabilistic Model for Vibration Signal Generation