GTAMP-DTA: Graph transformer combined with attention mechanism for drug-target binding affinity prediction

Chuangchuang Tian, Luping Wang, Zhiming Cui, Hongjie Wu

Abstract: Highlights•In order to enrich the spatial structure information of drugs and proteins, this study combined the drug and protein mapping conversion modules to obtain a richer feature vector. This approach effectively addresses the issue of information loss and mitigates the problem of irrational protein structure.•To model the complex non-covalent intermolecular interactions between drug atoms and amino acids, we employed an attention mechanism on the feature vectors, associating a vector with each drug atom and amino acid.•This study introduces self-supervised pre-trained embeddings to enhance protein/drug association signals. and integrates this advanced information into a unified framework to generate a richer drug-protein feature vector for predicting DTA.