Gene-Induced Drug Repurposing for Gastric Adenocarcinoma by Machine Learning

Gastric Cancer (GC) is one of the leading causes of cancer death worldwide. Drug repurposing plays a critical role in rapid drug discovery to resist the growth of tumor cells. In this project, we developed a computational drug repurposing pipeline (RepoGC) for gastric adenocarcinoma based on bulk-RNA transcriptomics profiles. With the gene expression profile of gastric cancer from the Gene Expression Omnibus (GEO), a total of 1004 differentially expressed genes were identified and used for the generation of potential drug targets. Over 10,000 drugs from DrugBank were used as drug candidates. The multi-omics networks, such as gene-gene interaction and gene-target interaction network, were constructed to discover the core genes and targets for repurposed drugs and assist drug ranking. We performed the drug-target interaction prediction by deep neural network to prioritize the repurposed drugs. Finally, a combined drug-target interaction score was generated to rank drug candidates. Among the top-ranked drugs, such as Ursolic acid, Geldanamycin, and Parthenolide, we explored their completed clinical trial evidence and studies in the field of gastric cancer to cross validate the drugs. The study presents an efficient computational method to integrate transcriptomics data for rapid drug repurposing. The project highlights the potential of RepoGC to identify new drugs for gastric adenocarcinoma.