Abstract
This study aims to identify core genes closely associated with the diagnosis and prognosis of colorectal cancer (CRC) using transcriptome-based meta-analysis approach and machine learning algorithms. Nine CRC datasets from the GEO database were integrated for differential gene expression analysis and WGCNA to identify key genes. Ninety-six combinations of machine learning algorithms were employed to further refine the selection of core genes and validate their diagnostic performance. Functional enrichment, molecular pathways, and associations with the immune microenvironment of core genes were analyzed using GSEA, CIBERSORT, and ssGSEA. Drug sensitivity predictions were performed to evaluate the impact of core genes on CRC drug response, and molecular docking simulations were used to identify candidate compounds targeting the core genes. A total of 26 core genes were identified, among which the high expression of the SACS gene was significantly associated with poor prognosis, advanced stage, and specific pathological subtypes in CRC patients. GSEA revealed that high SACS expression prominently activates cell cycle regulatory pathways and immune pathways while suppressing metabolic pathways. Furthermore, in vitro experiments demonstrated that SACS is highly expressed in CRC cells and that its knockdown significantly inhibits CRC cell proliferation, suggesting its functional role in tumor growth. Immune analysis showed that high SACS expression was positively correlated with activated NK cells but negatively correlated with Tregs and resting NK cells. Drug sensitivity analysis indicated that high SACS expression reduces sensitivity to oxaliplatin. Molecular docking identified coumestrol and quercetin as potential compounds targeting SACS. SACS promotes CRC progression by regulating cell cycle pathways, the immune microenvironment, and metabolic pathways. And it may serve as a potential therapeutic target for CRC.