DEAttentionDTA: protein–ligand binding affinity prediction based on dynamic embedding and self-attention

• DOI: 10.1093/bioinformatics/btae319
• Publication Date: 2024-06-20T00:45:57Z
• Authors: Xiying Chen; Jinsha Huang; Tianqiao Shen; Houjin Zhang; Li Xu; Min Yang; Xiaoman Xie; Yunjun Yan; Jinyong Yan
• Keywords: 0301 basic medicine; Original Paper; 03 medical and health sciences; Binding Sites; Proteins; Computational Biology; Neural Networks, Computer; Amino Acid Sequence; Ligands; Databases, Protein; Software; Protein Binding

Abstract Motivation Predicting protein–ligand binding affinity is crucial in new drug discovery and development. However, most existing models rely on acquiring 3D structures of elusive proteins. Combining amino acid sequences with ligand better highlighting active sites are also significant challenges. Results We propose an innovative neural network model called DEAttentionDTA, based dynamic word embeddings a self-attention mechanism, for predicting affinity. DEAttentionDTA takes the 1D sequence information proteins as input, including global features acids, local pocket site, linear representation molecule SMILE format. These three fed into word-embedding layer convolutional embedding encoding correlated through mechanism. The output prediction values generated using layer. compared various mainstream tools achieved significantly superior results same dataset. then assessed performance this p38 protein family. Availability implementation resource codes available at https://github.com/whatamazing1/DEAttentionDTA.