BridgeDPI:预测药物-蛋白质相互作用的新型图形神经网络,arXiv...

来自 : www.x-mol.com/paper/1356350813 发布时间：2021-03-25

Motivation: Exploring drug-protein interactions (DPIs) work as a pivotal stepin drug discovery. The fast expansion of available biological data enablescomputational methods effectively assist in experimental methods. Among them,deep learning methods extract features only from basic characteristics, such asprotein sequences, molecule structures. Others achieve significant improvementby learning from not only sequences/molecules but the protein-protein anddrug-drug associations (PPAs and DDAs). The PPAs and DDAs are generallyobtained by using computational methods. However, existing computationalmethods have some limitations, resulting in low-quality PPAs and DDAs thathamper the prediction performance. Therefore, we hope to develop a novelsupervised learning method to learn the PPAs and DDAs effectively and therebyimprove the prediction performance of the specific task of DPI. Results: Inthis research, we propose a novel deep learning framework, namely BridgeDPI.BridgeDPI introduces a class of nodes named hyper-nodes, which bridge differentproteins/drugs to work as PPAs and DDAs. The hyper-nodes can be supervisedlearned for the specific task of DPI since the whole process is an end-to-endlearning. Consequently, such a model would improve prediction performance ofDPI. In three real-world datasets, we further demonstrate that BridgeDPIoutperforms state-of-the-art methods. Moreover, ablation studies verify theeffectiveness of the hyper-nodes. Last, in an independent verification,BridgeDPI explores the candidate bindings among COVID-19\'s proteins and variousantiviral drugs. And the predictive results accord with the statement of theWorld Health Organization and Food and Drug Administration, showing thevalidity and reliability of BridgeDPI.