The paper presents a novel computational framework for predicting drug-protein interactions (DPIs) by integrating global and local features of nodes in the drug-protein bipartite graph. The framework employs pre-trained models to acquire initial features of drugs and proteins, uses MinHash and HyperLogLog scoring functions to estimate local features, and incorporates an energy-constrained diffusion mechanism into the transformer architecture to capture global features. The model's performance is evaluated on multiple datasets, including BindingDB, Davis, Enzyme, GPCR, and NR, demonstrating superior accuracy and reliability compared to existing models. The proposed model's ability to identify unknown DPIs is validated through molecular docking experiments, highlighting its potential for drug repurposing and personalized medicine research. The study contributes to the field by providing a comprehensive and efficient approach to DPI prediction, enhancing the understanding of drug mechanisms and facilitating the development of new treatments.The paper presents a novel computational framework for predicting drug-protein interactions (DPIs) by integrating global and local features of nodes in the drug-protein bipartite graph. The framework employs pre-trained models to acquire initial features of drugs and proteins, uses MinHash and HyperLogLog scoring functions to estimate local features, and incorporates an energy-constrained diffusion mechanism into the transformer architecture to capture global features. The model's performance is evaluated on multiple datasets, including BindingDB, Davis, Enzyme, GPCR, and NR, demonstrating superior accuracy and reliability compared to existing models. The proposed model's ability to identify unknown DPIs is validated through molecular docking experiments, highlighting its potential for drug repurposing and personalized medicine research. The study contributes to the field by providing a comprehensive and efficient approach to DPI prediction, enhancing the understanding of drug mechanisms and facilitating the development of new treatments.