Predicting Anticancer Drug Repurposing Candidates using Knowledge Graphs

Drug repurposing (DR) offers a promising and efficient alternative to traditional drug discovery by identifying new therapeutic applications for existing drugs, reducing the time and costs associated with development. This study introduces a novel framework that leverages a new hybrid knowledge graph integrating drug, disease, and protein interactions, combined with a dual-channel Convolutional Neural Network for drug-disease association prediction. The knowledge graph captures complex biological relationships through diverse biomedical data, while the neural network architecture enhances the model's ability to extract meaningful patterns. The framework demonstrates superior performance, achieving an AUC of ۰.۹۸۳۶ and AUPRC of ۰.۹۶۸۶, significantly outperforming state-of-the-art methods. To enhance the reliability of these predictions, molecular docking simulations were conducted, providing crucial biological validation. Integrating advanced machine learning with robust biological validation offers a promising avenue for accelerating drug discovery efforts and addressing critical unmet medical needs.

کلیدواژه ها:

Drug repurposing ، Dual_Channel Neural network ، Knowledge Graph ، Anticancer