TRIDENT: Tri-Modal Molecular Representation Learning with Taxonomic Annotations and Local Correspondence

2025-06-26

AI Summary

Existing molecular representation learning methods predominantly rely on SMILES strings while neglecting complementary textual descriptions and taxonomic annotations, leading to semantically impoverished representations. To address this, we propose the first trimodal molecular representation learning framework that jointly models SMILES, natural-language descriptions, and functional taxonomy annotations. Our method introduces two novel alignment mechanisms: (i) a global “volume alignment” objective that enforces distributional consistency across modality-specific embedding spaces, and (ii) a local “substructure–subtext” fine-grained alignment that bridges molecular substructures with corresponding descriptive phrases. A momentum-balancing strategy enables joint optimization of global and local objectives. The architecture employs three dedicated encoders and performs end-to-end, hierarchical functional semantic learning. Extensive evaluation across 11 molecular property prediction tasks demonstrates consistent and significant improvements over state-of-the-art methods, validating the critical role of integrating heterogeneous semantic information for robust molecular representation.

Technology Category

• Machine Learning: Multimodal Learning
• Natural Language Processing: Language Grounding & Multi-modal NLP
• Computer Vision: Representation Learning for Vision

Application Category

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

Abstract

Molecular property prediction aims to learn representations that map chemical structures to functional properties. While multimodal learning has emerged as a powerful paradigm to learn molecular representations, prior works have largely overlooked textual and taxonomic information of molecules for representation learning. We introduce TRIDENT, a novel framework that integrates molecular SMILES, textual descriptions, and taxonomic functional annotations to learn rich molecular representations. To achieve this, we curate a comprehensive dataset of molecule-text pairs with structured, multi-level functional annotations. Instead of relying on conventional contrastive loss, TRIDENT employs a volume-based alignment objective to jointly align tri-modal features at the global level, enabling soft, geometry-aware alignment across modalities. Additionally, TRIDENT introduces a novel local alignment objective that captures detailed relationships between molecular substructures and their corresponding sub-textual descriptions. A momentum-based mechanism dynamically balances global and local alignment, enabling the model to learn both broad functional semantics and fine-grained structure-function mappings. TRIDENT achieves state-of-the-art performance on 11 downstream tasks, demonstrating the value of combining SMILES, textual, and taxonomic functional annotations for molecular property prediction.

Problem

• Integrates SMILES, text, taxonomy for molecular representation
• Improves multimodal alignment with geometry-aware and local objectives
• Enhances molecular property prediction across 11 tasks

Innovation

• Integrates SMILES, text, and taxonomic annotations
• Uses volume-based alignment for tri-modal features
• Introduces local alignment for substructure-text relations