Deep learning-based virtual screening system for drug molecules

In the field of drug discovery, traditional virtual screening methods face challenges of being time-consuming, costly, and limited in accuracy.To address this, this study developed a deep learning-based virtual screening system for drug molecules.By automatically learning key molecular features through graph neural networks, the system overcomes the limitations of traditional methods that rely on manual feature extraction, thereby capturing more complex structural information.Testing on the public benchmark Directory of Useful Decoys: Enhanced (DUD-E) demonstrates that this system achieves an area under the curve (AUC) of 0.889 while significantly reducing screening time -approximately 80% faster than conventional methods.This research provides an efficient solution for rapidly and accurately identifying potential drug candidates from vast compound libraries, paving the way for accelerated drug development.