Molecule Graph Networks with Many-body Equivariant Interactions

2024-06-19

arXiv.org

AI Summary

Traditional message-passing neural networks (MPNNs) for molecular interaction prediction suffer from directional geometric information loss due to cancellation of reverse forces. To address this, we propose the Equivariant N-body Interaction Network (ENINet). ENINet is the first model to systematically establish both the necessity and generalizability of *l*=1 equivariant *N*-body interactions, thereby overcoming the inherent limitation of pairwise bond-vector modeling. Leveraging an SE(3)-equivariant graph neural network architecture, it integrates group representation-theoretic message passing, *l*=1 spherical harmonic vector features, and multi-body tensor-product interaction mechanisms—significantly enhancing symmetry-aware geometric reasoning. On benchmark datasets including QM9 and MD17, ENINet achieves substantial accuracy improvements across both scalar and tensor property prediction tasks.

Technology Category

• Machine Learning: Matrix & Tensor Methods
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning
• Natural Language Processing: Sentence-level Semantics, Textual Inference, etc.

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Web query analysis, representation and understanding

Abstract

Message passing neural networks have demonstrated significant efficacy in predicting molecular interactions. Introducing equivariant vectorial representations augments expressivity by capturing geometric data symmetries, thereby improving model accuracy. However, two-body bond vectors in opposition may cancel each other out during message passing, leading to the loss of directional information on their shared node. In this study, we develop Equivariant N-body Interaction Networks (ENINet) that explicitly integrates l = 1 equivariant many-body interactions to enhance directional symmetric information in the message passing scheme. We provided a mathematical analysis demonstrating the necessity of incorporating many-body equivariant interactions and generalized the formulation to $N$-body interactions. Experiments indicate that integrating many-body equivariant representations enhances prediction accuracy across diverse scalar and tensorial quantum chemical properties.

Problem

• Molecular Interaction Prediction
• Directional Information Loss
• Information Passing Neural Networks

Innovation

• ENINet
• Directional Information Preservation
• Complex Interaction Prediction