Abstract
Colorectal cancer (CRC) is highly heterogeneous, and the contribution of deubiquitination (DUB) programs to immune context and clinical outcome remains incompletely defined. We integrated transcriptomes from TCGA and GTEx (COAD/READ) with an external GEO cohort (GSE39582) to delineate DUB-associated features. Differential expression and univariate survival analyses yielded DUB-related, prognosis-associated genes, which were used for consensus clustering to derive CRC subtypes. Functional enrichment (GO/KEGG) and preranked GSEA characterized pathways; a STRING–Cytoscape–cytoHubba workflow identified hub genes; immune landscapes were profiled by CIBERSORT together with immune-checkpoint, immunogenic cell death (ICD) and HLA gene sets, and computational immunotherapy-related metrics (TIDE, IPS). Seventeen key DUB-related genes enriched in cell-cycle/DNA-damage response and ECM/EMT programs classified tumors into two robust subtypes with distinct overall survival, clinicopathologic profiles and transcriptional signatures. Sixty-six cluster-related DEGs formed an interaction network that highlighted nine ECM-centric hub genes with consistent prognostic value across TCGA and GEO. Subtypes differed markedly in immune-cell infiltration and in ICG/ICD/HLA expression, with clear separation of TIDE and IPS scores, indicating distinct computational immunotherapy-related signatures. These results define interpretable DUB-anchored immune subtypes and nominate hub genes as candidate biomarkers to support prognosis assessment and facilitate hypothesis-generating immunotherapy-related stratification. Prospective and functional validation is warranted, and this study should be interpreted as an exploratory analysis based on publicly available transcriptomic datasets. SUPPLEMENTARY INFORMATION: The online version contains supplementary material available at 10.1038/s41598-026-35271-5.