Cross-Domain Molecular Relational Learning: Leveraging Chemical Structure-Activity Analysis

2026-05-16

AI Summary

Existing molecular relation learning methods are often confined to a single data domain, limiting their ability to model cross-domain molecular interaction mechanisms. This work proposes DisTrans, a novel framework that integrates structure–activity analysis with cross-modal representation learning. By leveraging a gradient reversal strategy based on substructure topological discrepancies and a functional group semantic alignment mechanism, DisTrans jointly optimizes structural separability and semantic consistency. The approach synergistically combines domain-adversarial training, structure–semantic transfer discrepancy modeling, and cross-domain representation guidance. Evaluated under two representative cross-domain settings, DisTrans significantly outperforms 16 baseline methods and maintains robust performance even when inter-domain disparities are substantial.

Abstract

Recent advances in molecular representation integrates molecular topological and visual modalities, opening new avenues for precise Molecular Relational Learning (MRL). Existing MRL methods focus on intra-domain modeling, and their inherent domain-closed effect limits applicability to molecular science, particularly in elucidating cross-domain interaction mechanisms. Consequently, the imperative for Cross-Domain Molecular Relational Learning has become increasingly pressing. Benefiting from structure-activity analysis, we propose the Domain Adversarial Training Network with Structural-Semantic Transfer Discrepancy (DisTrans) to optimize cross-domain adaptive representation for molecular structures and visual images. 1) We employ the gradient reversal strategy based on substructure topological discrepancies between domains to learn the domain dependence of molecular structures. This strategy guides the model to adapt to the structural adjacency patterns in the target domain, generating domain-separable structural representations. 2) We apply the cross-domain representation guidance mechanism to align the functional-group semantic information between the source and target domains, learning cross-domain consistency information. The experimental results in two typical cross-domain strategies demonstrate that DisTrans outperforms 16 baseline methods, maintaining satisfactory performance even under pronounced inter-domain discrepancy.

Problem

• Cross-Domain
• Molecular Relational Learning
• Structure-Activity Analysis
• Domain Adaptation
• Molecular Representation

Innovation

• Cross-Domain Molecular Relational Learning
• Domain Adversarial Training
• Structural-Semantic Transfer Discrepancy
• Gradient Reversal Strategy
• Functional-Group Semantic Alignment