A three-gene signature (ABCE1, ATAD5, FUT4) for colorectal cancer prognosis: integrative analysis of transcriptomic, single-cell, and machine learning data.

Colorectal cancer (CRC) remains a leading cause of cancer-related death, highlighting an unmet need for robust, mechanistically grounded prognostic biomarkers. Through an integrative multi-omics approach, we identified and validated a novel three-gene signature for CRC. Utilizing paired tumor and adjacent tissues from 15 patients, public datasets (TCGA, GSE231559), and machine learning (LASSO, SVM-RFE), we derived a signature comprising ABCE1, ATAD5, and FUT4. These genes were consistently upregulated in tumors, as confirmed by qPCR, and are functionally linked to core oncogenic pathways: ABCE1 (protein synthesis/immune modulation), ATAD5 (DNA replication stress response), and FUT4 (cell adhesion/immune evasion). The signature demonstrated high diagnostic accuracy (AUC 0.85) and significant prognostic value, stratifying patients into distinct risk groups with divergent survival outcomes. Single-cell RNA-seq analysis localized expression to specific cellular compartments, while immune deconvolution revealed a correlated macrophage-dominated microenvironment. A clinically interpretable nomogram was developed with excellent calibration. This study establishes a compact, biologically coherent three-gene signature as a promising tissue-based molecular classifier for CRC, offering novel insights into the intertwined mechanisms of metabolism, genomic instability, and immune suppression, and presenting potential targets for therapeutic development.