DTIAM: a unified framework for predicting drug-target interactions, binding affinities and drug mechanisms

• DOI: 10.1038/s41467-025-57828-0
• Publication Date: 2025-03-15T10:44:28Z
• Authors: Zhangli Lu; Guoqiang Song; Huimin Zhu; Chuqi Lei; Xinliang Sun; Kaili Wang; Libo Qin; Yafei Chen; Jing Tang; Min Li
• Keywords: Affinities; Binding affinities; Drug target

Accurate and robust prediction of drug-target interactions (DTIs) plays a vital role in drug discovery but remains challenging due to limited labeled data, cold start problems, insufficient understanding mechanisms action (MoA). Distinguishing activation inhibition is particularly critical clinical applications. Here, we propose DTIAM, unified framework for predicting interactions, binding affinities, activation/inhibition between drugs targets. DTIAM learns target representations from large amounts label-free data through self-supervised pre-training, which accurately extracts their substructure contextual information, thus benefits the downstream based on these representations. achieves substantial performance improvement over other state-of-the-art methods all tasks, scenario. Moreover, independent validation demonstrates strong generalization ability DTIAM. All results suggest that can provide practically useful tool novel DTIs further distinguishing MoA candidate drugs. Accurately are discovery. The authors here affinities mechanisms.