MMPG: MoE-based Adaptive Multi-Perspective Graph Fusion for Protein Representation Learning

2026-01-15

AI Summary

Existing graph neural network–based approaches for protein representation typically rely on a single perspective to construct residue interaction graphs, limiting their ability to comprehensively capture the structural and functional properties of proteins. To address this limitation, this work proposes the first multi-view graph fusion framework based on a Mixture-of-Experts (MoE) mechanism, which constructs protein graphs from three complementary perspectives—physical, chemical, and geometric—and dynamically models both view-specific features and their synergistic relationships through adaptive routing. This enables hierarchical interactions spanning from individual views to cross-view integration and ultimately to global consensus. The proposed method significantly outperforms current state-of-the-art approaches across four downstream protein tasks, demonstrating the effectiveness of multi-view fusion and the MoE architecture in enhancing protein representation quality.

Technology Category

• Machine Learning: Mixture of Experts (MoE)
• Computer Vision: Multi-modal Vision
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches 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: Retrieval-Augmented Generation (RAG) and multi-modal RAG

Abstract

Graph Neural Networks (GNNs) have been widely adopted for Protein Representation Learning (PRL), as residue interaction networks can be naturally represented as graphs. Current GNN-based PRL methods typically rely on single-perspective graph construction strategies, which capture partial properties of residue interactions, resulting in incomplete protein representations. To address this limitation, we propose MMPG, a framework that constructs protein graphs from multiple perspectives and adaptively fuses them via Mixture of Experts (MoE) for PRL. MMPG constructs graphs from physical, chemical, and geometric perspectives to characterize different properties of residue interactions. To capture both perspective-specific features and their synergies, we develop an MoE module, which dynamically routes perspectives to specialized experts, where experts learn intrinsic features and cross-perspective interactions. We quantitatively verify that MoE automatically specializes experts in modeling distinct levels of interaction from individual representations, to pairwise inter-perspective synergies, and ultimately to a global consensus across all perspectives. Through integrating this multi-level information, MMPG produces superior protein representations and achieves advanced performance on four different downstream protein tasks.

Problem

• Protein Representation Learning
• Graph Neural Networks
• Multi-Perspective Graph
• Residue Interaction
• Incomplete Representation

Innovation

• Mixture of Experts
• Multi-perspective Graph Fusion
• Protein Representation Learning
• Graph Neural Networks
• Adaptive Fusion