An integrative drug discovery strategy using real-world data, in silico modeling, and network pharmacology

This study evaluated the utility of an integrated drug discovery strategy that combines three emerging data-driven approaches: real-world data analysis, in silico screening, and network pharmacology. First, transcriptomic data from public gene expression databases and adverse event reports were analyzed to address myocarditis induced by immune checkpoint inhibitors. The findings suggested a preventive effect of non-steroidal anti-inflammatory drugs, particularly those targeting the arachidonic acid metabolism pathway. Second, to identify therapeutic options for trastuzumab-resistant HER2-positive breast cancer, a cheminformatics approach was applied. A machine learning classification model and structure-based docking simulations enabled efficient in silico screening of approved drugs, identifying novel YES1 kinase inhibitors. Third, network-based analysis evaluated the topological distance between disease-associated gene modules and statin-induced gene modules in drug-induced peripheral neuropathy. This analysis indicated that certain statins may protect against drug-induced peripheral neuropathy through modulation of shared targets and neurodegenerative pathways. These findings demonstrate that integrating heterogeneous data modalities-from transcriptomics and chemical structure to protein-protein interaction networks and real-world clinical observations-can enable the discovery of repositioning candidates and risk-mitigating therapies. The study highlights the potential of multi-layered, data-driven strategies in constructing translational drug discovery frameworks aimed at both efficacy and safety.