HyperAttentionDTI: improving drug–protein interaction prediction by sequence-based deep learning with attention mechanism

• DOI: 10.1093/bioinformatics/btab715
• Publication Date: 2021-10-13T19:30:18Z
• Authors: Qichang Zhao; Haochen Zhao; Kai Zheng; Jianxin Wang
• Keywords: Benchmark (surveying); Feature (linguistics); Drug repositioning; Sequence (biology)

Abstract Motivation Identifying drug–target interactions (DTIs) is a crucial step in drug repurposing and discovery. Accurately identifying DTIs silico can significantly shorten development time reduce costs. Recently, many sequence-based methods are proposed for DTI prediction improve performance by introducing the attention mechanism. However, these only model single non-covalent inter-molecular among drugs proteins ignore complex interaction between atoms amino acids. Results In this article, we propose an end-to-end bio-inspired based on convolutional neural network (CNN) mechanism, named HyperAttentionDTI, predicting DTIs. We use deep CNNs to learn feature matrices of proteins. To acids, utilize mechanism assign vector each atom or acid. evaluate HpyerAttentionDTI three benchmark datasets results show that our achieves improved compared with state-of-the-art baselines. Moreover, case study human Gamma-aminobutyric acid receptors confirm be used as powerful tool predict Availability implementation The codes available at https://github.com/zhaoqichang/HpyerAttentionDTI https://zenodo.org/record/5039589. Supplementary information data Bioinformatics online.