JMIR Publications

ABSTRACT

Background:

Drug-target interaction prediction is crucial in drug repositioning, as it can significantly reduce research and development costs and shorten the development cycle. Most existing deep learning-based approaches employ graph neural networks for drug-target interaction prediction. However, these approaches still face limitations in capturing complex biochemical features, integrating multi-level information, and interpreting model interpretability.

Objective:

This study proposes a heterogeneous network model based on multi-view path aggregation, aiming to predict interactions between drugs and targets.

Methods:

This study employs a molecular attention Transformer to extract 3D conformation features from the chemical structures of drugs, and utilizes Prot-T5, a protein-specific large language model, to deeply explore key semantic features in protein sequences. By integrating drugs, proteins, diseases, and side effects from multi-source heterogeneous data, we construct a heterogeneous graph model to systematically characterize multi-dimensional associations between biological entities. On this foundation, a meta-path aggregation mechanism is proposed, which dynamically integrates information from both feature views and biological network relationship views. This mechanism effectively learns potential interaction patterns between biological entities and provides a more comprehensive representation of the complex relationships in the heterogeneous graph. It enhances the model's ability to capture sophisticated, context-dependent relationships in biological networks. Furthermore, we integrate multi-scale features of drugs and proteins within the heterogeneous network, significantly improving the prediction accuracy of drug-target interactions, and enhancing the model's interpretability and generalization ability.

Results:

In the drug-target interaction prediction task, the proposed model achieves an AUPR of 0.901 and an AUROC of 0.967, representing improvements of 1.7% and 0.9%, respectively, over the baseline methods. Furthermore, a case study on the KCNH2 target demonstrates that the proposed model successfully predicts 38 out of 53 candidate drugs as having interactions, which further validates its reliability and practicality in real-world scenarios.

Conclusions:

The proposed model shows marked superiority over baseline methods, highlighting the importance of integrating heterogeneous information with biological knowledge in drug-target interactions prediction. Clinical Trial: This study does not involve any clinical trials. All data were obtained from publicly available datasets, including DrugBank version 3.0, HPRD Release 9, Comparative Toxicogenomics Database, and SIDER Release 2.