Construction of a gene-based diagnostic model for prostate cancer and AI-guided drug prediction coupled with molecular docking

To construct a gene-based diagnostic model for prostate cancer and to predict potential therapeutics using artificial intelligence (AI) based on datasets from the GEO database. Differential gene expression analysis was performed between tumor and normal samples using expression matrices. Weighted gene co-expression network analysis (WGCNA) was applied to the corrected expression data to identify modules of co-expressed genes. The intersection of significantly differentially expressed genes and WGCNA-derived genes was subjected to Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses. A protein–protein interaction (PPI) network was constructed from the intersecting genes, and the top five hub genes, ranked by degree of connectivity, were used to develop and validate a diagnostic model for prostate cancer. Finally, AI-driven drug prediction was conducted using the gene expression profiles. A total of 516 differentially expressed genes and 3,392 WGCNA-derived co-expressed genes were identified, yielding 465 intersecting genes associated with prostate cancer. GO enrichment analysis revealed significant involvement in oxygen level response, collagen-containing extracellular matrix, and enzyme inhibitor activity. KEGG pathway analysis indicated enrichment in cancer-related pathways, including prostate cancer, hepatocellular carcinoma, drug metabolism (CYP450), Wnt signaling, AMPK signaling, and oxidative stress–related pathways. In the PPI network, the top five hub genes—ANXA2, GJA1, PIK3R1, CAV1, and MYC—demonstrated strong diagnostic performance in the prostate cancer prediction model. AI-based prediction identified AGI-5198 as a candidate therapeutic agent. ANXA2, GJA1, PIK3R1, CAV1, and MYC represent key predictive genes for prostate cancer diagnosis, while AGI-5198 emerges as a promising AI-predicted therapeutic candidate.