Integrating single-cell RNA sequencing, WGCNA, and machine learning to identify key biomarkers in hepatocellular carcinoma.

Abstract

The microarray and single-cell RNA-sequencing (scRNA-seq) datasets of hepatocellular carcinoma (HCC) were downloaded from the Gene Expression Omnibus (GEO) database. Differential expression analysis and weighted gene co-expression network analysis (WGCNA) were used to identify HCC-related biomarkers. Based on an analysis of scRNA-seq data, several marker genes expressed on tumor cells have been identified. Three machine-learning algorithms were used to identify shared diagnostic genes. Furthermore, logistic regression analysis was conducted to re-evaluate and identify essential biomarkers, which were then employed to develop a diagnostic prediction model. Additionally, AutoDockTools were used for molecular docking to investigate the association between the most sensitive drug and the core proteins. 44 genes were obtained by intersecting the WGCNA results, marker genes from scRNA-seq data, and up-regulated DEGs. Three machine-learning algorithms refined CDKN3, PPIA, PRC1, GMNN, and CENPW as hub biomarkers. GMNN and PRC1 were further selected by logistic regression analysis to build a nomogram. The molecular docking results showed that the drug NPK76-II-72-1 had a good binding ability with the GMNN and PRC1 proteins. The results highlighted CDKN3, PPIA, PRC1, GMNN, and CENPW as potential detection biomarkers for HCC patients. Our research offers novel insights into the diagnosis and treatment of HCC.