Topology-Aware Multiscale Mixture of Experts for Efficient Molecular Property Prediction

2026-01-19

AI Summary

Current 3D molecular graph neural networks are limited by fixed distance cutoffs and a constrained number of neighbors, hindering their ability to effectively model multiscale non-covalent interactions. To address this, this work proposes the Multiscale Interaction Mixture-of-Experts (MI-MoE) framework, which dynamically allocates modeling tasks for short-, medium-, and long-range interactions via a topology-aware gating mechanism. The approach integrates topological features—such as persistent homology—into a gating encoder and employs a plug-and-play ensemble of expert modules, ensuring compatibility with mainstream 3D molecular backbone architectures. Evaluated across multiple benchmarks for molecular and polymer property prediction, MI-MoE consistently enhances both regression and classification performance as a drop-in module, demonstrating its effectiveness and generality in multiscale geometric modeling.

Technology Category

• Machine Learning: Mixture of Experts (MoE)
• Computer Vision: Multi-modal Vision
• Data Mining & Knowledge Management: Graph Mining, Social Network Analysis & Community

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
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Many molecular properties depend on 3D geometry, where non-covalent interactions, stereochemical effects, and medium- to long-range forces are determined by spatial distances and angles that cannot be uniquely captured by a 2D bond graph. Yet most 3D molecular graph neural networks still rely on globally fixed neighborhood heuristics, typically defined by distance cutoffs and maximum neighbor limits, to define local message-passing neighborhoods, leading to rigid, data-agnostic interaction budgets. We propose Multiscale Interaction Mixture of Experts (MI-MoE) to adapt interaction modeling across geometric regimes. Our contributions are threefold: (1) we introduce a distance-cutoff expert ensemble that explicitly captures short-, mid-, and long-range interactions without committing to a single cutoff; (2) we design a topological gating encoder that routes inputs to experts using filtration-based descriptors, including persistent homology features, summarizing how connectivity evolves across radii; and (3) we show that MI-MoE is a plug-in module that consistently improves multiple strong 3D molecular backbones across diverse molecular and polymer property prediction benchmark datasets, covering both regression and classification tasks. These results highlight topology-aware multiscale routing as an effective principle for 3D molecular graph learning.

Problem

• molecular property prediction
• 3D molecular graph
• non-covalent interactions
• topology-aware modeling
• multiscale interaction

Innovation

• Mixture of Experts
• Persistent Homology
• 3D Molecular Graph Neural Networks
• Topology-Aware Routing
• Multiscale Interaction Modeling