Pre-training Graph Neural Networks on 2D and 3D Molecular Structures by using Multi-View Conditional Information Bottleneck

2025-11-23

AI Summary

Existing molecular graph pretraining struggles to simultaneously ensure cross-view (2D/3D) semantic consistency and alignment of functionally critical substructures. To address this, we propose the Multi-View Conditional Information Bottleneck (MVCIB) framework, which achieves fine-grained cross-view alignment via context-guided representation learning and a functional-group-driven substructure anchoring mechanism. MVCIB is the first method to attain geometric discriminability at the 3D Weisfeiler–Lehman hierarchy level. It integrates ego-network modeling with cross-view attention to jointly optimize 2D topological and 3D geometric representations. Evaluated on four molecular property prediction benchmarks, MVCIB significantly outperforms state-of-the-art methods—particularly excelling at distinguishing stereoisomers with identical 2D graphs but distinct 3D conformations. The framework delivers both superior predictive performance and enhanced interpretability through principled substructure grounding.

Technology Category

• Machine Learning: Multi-instance/Multi-view Learning
• Computer Vision: Multi-modal Vision
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning

Application Category

• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning 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 learning to rank, online learning, and counterfactual learning for ranking

Abstract

Recent pre-training strategies for molecular graphs have attempted to use 2D and 3D molecular views as both inputs and self-supervised signals, primarily aligning graph-level representations. However, existing studies remain limited in addressing two main challenges of multi-view molecular learning: (1) discovering shared information between two views while diminishing view-specific information and (2) identifying and aligning important substructures, e.g., functional groups, which are crucial for enhancing cross-view consistency and model expressiveness. To solve these challenges, we propose a Multi-View Conditional Information Bottleneck framework, called MVCIB, for pre-training graph neural networks on 2D and 3D molecular structures in a self-supervised setting. Our idea is to discover the shared information while minimizing irrelevant features from each view under the MVCIB principle, which uses one view as a contextual condition to guide the representation learning of its counterpart. To enhance semantic and structural consistency across views, we utilize key substructures, e.g., functional groups and ego-networks, as anchors between the two views. Then, we propose a cross-attention mechanism that captures fine-grained correlations between the substructures to achieve subgraph alignment across views. Extensive experiments in four molecular domains demonstrated that MVCIB consistently outperforms baselines in both predictive performance and interpretability. Moreover, MVCIB achieved the 3d Weisfeiler-Lehman expressiveness power to distinguish not only non-isomorphic graphs but also different 3D geometries that share identical 2D connectivity, such as isomers.

Problem

• Discovering shared molecular information while minimizing view-specific irrelevant features
• Identifying and aligning important substructures like functional groups across views
• Enhancing cross-view consistency and model expressiveness for molecular structures

Innovation

• Multi-view conditional information bottleneck for molecular graphs
• Key substructures as anchors for cross-view consistency
• Cross-attention mechanism for fine-grained subgraph alignment