Drug repurposing offers a cost-effective strategy to identify new applications for existing medications, leveraging established safety profiles to accelerate therapeutic development. Advances in computational biology and large-scale multi-omics data enable systematic identification of novel therapeutic opportunities, addressing unmet medical needs and advancing precision medicine. This study employs a multiplex network integrating 10 literature-based layers from the HumanNet database and 320 data-driven predictive expression networks derived from single-cell RNA sequencing and bulk transcriptomic data. Constructed using the iRF-LOOP algorithm, requiring over 500,000 compute hours on the Frontier supercomputer, this multiplex provides a framework for analyzing gene functions across diverse biological contexts.We applied the Random Walk with Restart algorithm to compute embeddings for 52,722 genes, quantifying their topological relevance within the network. Drug-gene interactions from DrugBank and disease-gene associations from UKBiobank GWAS were mapped to these embeddings, linking therapeutic agents to potential targets and revealing biomarkers. A case study on glucagon-like peptide 1 receptor (GLP-1R) agonists, initially developed for type 2 diabetes, identified genes topologically connected to GLP-1R (TMPRSS2, PNPLA3, DHX37, ZNF91, DTHD1, and IRX3) and associated diseases. This study demonstrates the power of multiplex networks and supercomputing in uncovering connections between genes, drugs, and diseases, offering insights into therapeutic discovery.