Graph Network-Based Analysis of Disease-Gene-Drug Associations: Zero-Shot Disease-Drug Prediction and Analysis Strategies

ZGNT, an innovative, novel workflow for zero-shot learning in drug repurposing that leverages meta-path graph neural network transform- ers. This method infers disease-drug relationships indirectly, utilizing disease-gene associations and gene-drug interactions via acting genes. It also generates a TOP drug-TOP gene linkage map, providing clini- cians with a visual tool to assess the plausibility of the suggested drugs before advancing to clinical trials. Experimental results demonstrate that, among the top 1% of disease-related drug candidates identified by ZGNT, the proportion of usable drugs is ***, surpassing four base- line methods. Furthermore, the accuracy of interaction link prediction achieves ***, also surpassing these benchmarks. Ultimatley, by inte- grating data acquisition, model design, and application workflows, our approach effectively identifies potential drug targets, offering new in- sights into therapeutic strategies for disease management. This paper thoroughly details the methodology and validates its effectiveness across various diseases, highlighting its robustness and practical value.