Rethinking Drug-Drug Interaction Modeling as Generalizable Relation Learning

2026-01-22

AI Summary

Existing drug–drug interaction (DDI) prediction methods perform well on standard benchmarks but exhibit limited generalization in real-world scenarios involving unseen drugs or sparse annotations. This work reframes DDI prediction as a relation-centric learning problem and introduces GenRel-DDI, a novel framework that leverages a relation-level abstraction mechanism to learn interaction representations independent of specific drug identities. By capturing transferable interaction patterns, the approach overcomes the limitations of conventional molecule-centric paradigms. Extensive experiments demonstrate that GenRel-DDI significantly outperforms current state-of-the-art models across multiple benchmarks, with particularly substantial gains under strict entity-split settings where training and test drugs are disjoint. These results validate the efficacy of relation-centric learning in enhancing the generalization capability of DDI prediction systems.

Technology Category

• Reasoning under Uncertainty: Relational Probabilistic Models
• Knowledge Representation and Reasoning: Diagnosis and Abductive Reasoning
• Machine Learning: Statistical Relational/Logic Learning

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Drug-drug interaction (DDI) prediction is central to drug discovery and clinical development, particularly in the context of increasingly prevalent polypharmacy. Although existing computational methods achieve strong performance on standard benchmarks, they often fail to generalize to realistic deployment scenarios, where most candidate drug pairs involve previously unseen drugs and validated interactions are scarce. We demonstrate that proximity in the embedding spaces of prevailing molecule-centric DDI models does not reliably correspond to interaction labels, and that simply scaling up model capacity therefore fails to improve generalization. To address these limitations, we propose GenRel-DDI, a generalizable relation learning framework that reformulates DDI prediction as a relation-centric learning problem, in which interaction representations are learned independently of drug identities. This relation-level abstraction enables the capture of transferable interaction patterns that generalize to unseen drugs and novel drug pairs. Extensive experiments across multiple benchmark demonstrate that GenRel-DDI consistently and significantly outperforms state-of-the-art methods, with particularly large gains on strict entity-disjoint evaluations, highlighting the effectiveness and practical utility of relation learning for robust DDI prediction. The code is available at https://github.com/SZU-ADDG/GenRel-DDI.

Problem

• drug-drug interaction
• generalization
• unseen drugs
• relation learning
• polypharmacy

Innovation

• relation learning
• drug-drug interaction
• generalization
• entity-disjoint evaluation
• transferable interaction patterns