Directional Message Passing for Molecular Graphs

2020/03/06 by Johannes Gasteiger, Gasteiger, Johannes, Janek Groß +3 · 48 citations
Biochemistry, Genetics and Molecular Biology · Computer Science · Materials Science · #Computational Drug Discovery Methods #Computational Physics (physics.comp-ph) #FOS: Computer and information sciences #FOS: Physical sciences #Machine Learning (cs.LG) #Machine Learning (stat.ML) #Machine Learning in Bioinformatics #Machine Learning in Materials Science

paper · pdf · doi:10.48550/arxiv.2003.03123

openalex publication_date 2020/03/06 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

Graph neural networks have recently achieved great successes in predicting quantum mechanical properties of molecules. These models represent a molecule as a graph using only the distance between atoms (nodes). They do not, however, consider the spatial direction from one atom to another, despite directional information playing a central role in empirical potentials for molecules, e.g. in angular potentials. To alleviate this limitation we propose directional message passing, in which we embed the messages passed between atoms instead of the atoms themselves. Each message is associated with a direction in coordinate space. These directional message embeddings are rotationally equivariant since the associated directions rotate with the molecule. We propose a message passing scheme analogous to belief propagation, which uses the directional information by transforming messages based on the angle between them. Additionally, we use spherical Bessel functions and spherical harmonics to construct theoretically well-founded, orthogonal representations that achieve better performance than the currently prevalent Gaussian radial basis representations while using fewer than 1/4 of the parameters. We leverage these innovations to construct the directional message passing neural network (DimeNet). DimeNet outperforms previous GNNs on average by 76% on MD17 and by 31% on QM9. Our implementation is available online.

Cited by

• Faster Molecular Dynamics with Neural Network Potentials via Distilled Multiple Time-Stepping and Nonconservative Forces
• mCGCNN: A Dual-Stream Crystal Graph Convolutional Neural Network for the Efficient Prediction of Magnetic Properties of Crystalline Materials
• QE-Catalytic: A Graph-Language Multimodal Base Model for Relaxed-Energy Prediction in Catalytic Adsorption
• ChemATP: A Training-Free Chemical Reasoning Framework for Large Language Models
• Infusing Experimental Reality into Complex Many-Body Hamiltonians: The Observable-Constrained Variational Framework (OCVF)
• Drawback of Enforcing Equivariance and its Compensation via the Lens of Expressive Power
• Mapping Still Matters: Coarse-Graining with Machine Learning Potentials
• Learning Thermoelectric Transport from Crystal Structures via Multiscale Graph Neural Network
• QuantumCanvas: A Multimodal Benchmark for Visual Learning of Atomic Interactions
• Beyond Atoms: Evaluating Electron Density Representation for 3D Molecular Learning
• Equivariant Atomic and Lattice Modeling Using Geometric Deep Learning for Crystal Structure Optimization
• Toward bilipshiz geometric models
• Compact Artificial Neural Network Models for Predicting Protein Residue -- RNA Base Binding
• Magnitude-Modulated Equivariant Adapter for Parameter-Efficient Fine-Tuning of Equivariant Graph Neural Networks
• Design, Assessment, and Application of Machine Learning Potential Energy Surfaces
• Knowledge Distillation of a Protein Language Model Yields a Foundational Implicit Solvent Model
• Learning 3D Anisotropic Noise Distributions Improves Molecular Force Field Modeling
• Structure-Aware Fusion with Progressive Injection for Multimodal Molecular Representation Learning
• Layer-to-Layer Knowledge Mixing in Graph Neural Network for Chemical Property Prediction
• 3D-GSRD: 3D Molecular Graph Auto-Encoder with Selective Re-mask Decoding
• SimPoly: Simulation of Polymers with Machine Learning Force Fields Derived from First Principles
• Universally Invariant Learning in Equivariant GNNs
• Scaling Laws and Symmetry, Evidence from Neural Force Fields
• MolGA: Molecular Graph Adaptation with Pre-trained 2D Graph Encoder
• When Does Global Attention Help? A Unified Empirical Study on Atomistic Graph Learning
• Analyzing the Effect of Embedding Norms and Singular Values to Oversmoothing in Graph Neural Networks
• Flow-Matching Based Refiner for Molecular Conformer Generation
• Efficient E(3)-equivariant framework for universal charge density prediction
• Equivariance by Local Canonicalization: A Matter of Representation
• MCGM: Multi-stage Clustered Global Modeling for Long-range Interactions in Molecules
• Beyond Structure: Invariant Crystal Property Prediction with Pseudo-Particle Ray Diffraction
• Learning Inter-Atomic Potentials without Explicit Equivariance
• Shoot from the HIP: Hessian Interatomic Potentials without derivatives
• Chem World: A Large-Scale Benchmark and Physics-Informed Framework for Trustworthy Chemical Property Prediction
• Graph atomic cluster expansion for foundational machine learning interatomic potentials
• Spectral/Spatial Tensor Atomic Cluster Expansion with Universal Embeddings in Cartesian Space
• DEQuify your force field: More efficient simulations using deep equilibrium models
• Bayesian Prior Construction for Uncertainty Quantification in First-Principles Statistical Mechanics
• Global properties of the energy landscape: a testing and training arena for machine learned potentials
• Node-Equivariant Message Passing for Efficient and Accurate Machine Learning Interatomic Potentials
• What Expressivity Theory Misses: Message Passing Complexity for GNNs
• Data-driven particle dynamics: Structure-preserving coarse-graining for emergent behavior in non-equilibrium systems
• Recent advances and applications of deep learning methods in materials science
• Chi-Geometry: A Library for Benchmarking Chirality Prediction of GNNs
• Learning Potential Energy Surfaces of Hydrogen Atom Transfer Reactions in Peptides
• Rotational equivariant graph neural networks via local eigenbasis transformations
• Gradient-based grand canonical optimization enabled by graph neural networks with fractional atomic existence
• Learning Long-Range Representations with Equivariant Messages

Related

• Directional Message Passing on Molecular Graphs via Synthetic Coordinates
• GemNet: Universal Directional Graph Neural Networks for Molecules
• Equivariant Graph Attention Networks for Molecular Property Prediction
• ComENet: Towards Complete and Efficient Message Passing for 3D Molecular Graphs
• Neural Message Passing for Quantum Chemistry