1. A general hypergraph learning algorithm for drug multi-task predictions in micro-to-macro biomedical networks.
- Author
-
Jin, Shuting, Hong, Yue, Zeng, Li, Jiang, Yinghui, Lin, Yuan, Wei, Leyi, Yu, Zhuohang, Zeng, Xiangxiang, and Liu, Xiangrong
- Subjects
- *
MACHINE learning , *DEEP learning , *DRUG discovery , *MOLECULAR interactions , *FUNCTIONAL groups , *CHEMICAL structure - Abstract
The powerful combination of large-scale drug-related interaction networks and deep learning provides new opportunities for accelerating the process of drug discovery. However, chemical structures that play an important role in drug properties and high-order relations that involve a greater number of nodes are not tackled in current biomedical networks. In this study, we present a general hypergraph learning framework, which introduces Drug-Substructures relationship into Molecular interaction Networks to construct the micro-to-macro drug centric heterogeneous network (DSMN), and develop a multi-branches HyperGraph learning model, called HGDrug, for Drug multi-task predictions. HGDrug achieves highly accurate and robust predictions on 4 benchmark tasks (drug-drug, drug-target, drug-disease, and drug-side-effect interactions), outperforming 8 state-of-the-art task specific models and 6 general-purpose conventional models. Experiments analysis verifies the effectiveness and rationality of the HGDrug model architecture as well as the multi-branches setup, and demonstrates that HGDrug is able to capture the relations between drugs associated with the same functional groups. In addition, our proposed drug-substructure interaction networks can help improve the performance of existing network models for drug-related prediction tasks. Author summary: Drugs containing the same functional groups may have similar pharmacochemical properties. However, how to effectively combine chemical information of drugs from molecular fragments containing functional groups into the biomolecular network is challenging and rarely explored. we decompose drugs' SMILES string and construct a drug-centric heterogeneous network that integrates drug substructure and molecular interactions information. Based on the heterogeneous network, we proposed an end-to-end hypergraph attention network framework for the drug multi-task predictions, termed as HGDrug. The efficiency and generalization of the proposed HGDrug have been demonstrated by the state-of-the-art performance in four drug-related interaction predictions tasks with huge improvement compared to previous general-purpose classical models and task-specific models. In addition, HGDrug can effectively identify potential drug-related interactions and the drug-sub-structure networks are able to help to improve the performance of other GNN models. These conclusions present important insights on how to introduce the drug' substructure information for multiple drug-related interactions tasks on biomedical networks. In summary, HGDrug offers a general and powerful tool for the identification of drug-related interactions by constructing the micro-to-macro drug-centric heterogeneous network. [ABSTRACT FROM AUTHOR]
- Published
- 2023
- Full Text
- View/download PDF