Abstract
Hepatocellular carcinoma continues to be a predominant contributor to oncological fatalities, characterized by restricted treatment alternatives. Although berberine exhibits anti-neoplastic capabilities, the underlying molecular pathways in hepatic malignancy require clarification. A comprehensive computational framework was established, incorporating transcriptomic data analysis, multiple machine learning methodologies, weighted gene co-expression network analysis (WGCNA), and molecular simulation techniques to elucidate berberine's therapeutic pathways. Transcriptomic datasets from the Cancer Genome Atlas (TCGA) underwent examination to detect differentially expressed genes (DEGs). Ten machine learning methodologies screened critical targets, subsequently validated through molecular docking and 100 ns molecular dynamics simulations. Transcriptomic examination revealed 531 DEGs (341 exhibiting upregulation, 190 demonstrating downregulation) alongside 173 putative berberine interaction targets, yielding 17 intersecting candidates. Machine learning approaches consistently recognized AURKA and CDK1 as principal targets, subsequently confirmed by WGCNA as central genes. Elevated expression of both targets demonstrated correlation with unfavorable survival outcomes (p < 0.05). Computational docking analysis demonstrated robust binding interactions (AURKA: -8.2 kcal/mol; CDK1: -8.4 kcal/mol), with interaction stability validated through molecular dynamics simulations. Functional enrichment analysis unveiled targeting of cell cycle modulation, chromosome segregation, and p53 signaling networks. Berberine manifests anti-hepatocellular carcinoma activities primarily via coordinated targeting of AURKA and CDK1, essential cell cycle modulators. These discoveries provide molecular insights supporting berberine's potential as adjunctive hepatic cancer therapy.