Abstract

Background: The aggressiveness of liver cancer, especially hepatocellular carcinoma (HCC), and the lack of early diagnosis make liver cancer a significant global health issue. This study aims to analyse transcriptomic data from The Cancer Genome Atlas (TCGA-LIHC) project to identify patterns of differential gene expression between tumor and normal liver tissues. Materials and Methods: The analysis focused on the 20 most variable genes using statistical visualization methods such as boxplots, density plots, and heatmaps. Principal component analysis (PCA) and Random Forest classification models were employed to detect expression differences and assess predictive capabilities. Results: The findings revealed distinct expression patterns, with some genes (e.g., GLUL, FTL, APOA2) upregulated in tumor samples, while others (e.g., ALB, APOA1, ALDOB, HP, FGA) were downregulated, reflecting cancer-specific transcriptional activity. Strong gene–gene correlations suggested the presence of co-regulated modules potentially involved in oncogenic pathways. PCA demonstrated a clear separation between tumor and normal tissues, and the Random Forest model achieved high classification accuracy with a low error rate, particularly in identifying tumor tissues. Functional enrichment analysis further revealed significant involvement of coagulation, complement, and lipid metabolism pathways, providing biological context for the expression differences. External validation using an independent GEO cohort (GSE14520) confirmed the robustness of the 20-gene panel, with elastic-net achieving an AUC of 0.97 and Random Forest reaching an AUC of 0.9998, supporting reproducibility across platforms. Conclusion: These results support the potential of expression-based biomarkers in distinguishing liver cancer states and lay the groundwork for developing diagnostic or therapeutic strategies. Although further validation in independent cohorts and experimental settings is necessary, this study highlights the value of integrative bioinformatics in uncovering molecular signatures critical to advancing liver cancer research and clinical management.