Identification of the clinical prediction model and biomarkers based on chromatin regulators in colon cancer by integrated analysis of bulk- and single-cell RNA sequencing data

Abstract

Abstract Background Chromatin regulators (CRs) are implicated in the development of cancer, but a comprehensive investigation of their role in colon adenocarcinoma (COAD) is inadequate. This research's objective is to conduct a complete examination of CRs in COAD.Methods We obtained data from TCGA and GEO databases. WGCNA screened tumor-associated CRs. Lasso-cox regression was used to construct the model and to screen key CRs together with SVM, the univariate cox regression. We used single-cell data to explore expression of CRs in cells and their communication. Immune infiltration, immune checkpoints, mutation, methylation, and drug sensitivity analyses were performed. Gene expression was verified by qRT-PCR. Pan-cancer analysis was used to explore the importance of hub CRs.Results We finally obtained 32 tumor-associated CRs. The prognostic model was constructed based on RCOR2, PPARGC1A, PKM, RAC3, PHF19, MYBBP1A, ORC1, and EYA2 by the Lasso-cox regression. Single cell data revealed that the model was immune-related. Substantial differences existed between the high-risk and low-risk cohorts in the clinical features, tumor microenvironment, and drug sensitivity. Combined with machine learning, PKM is perhaps the most critical gene in CRs. Pan-cancer analysis showed that PKM plays a role in the prognosis of cancers.Conclusions We developed a prognostic model for COAD dependent on CRs. Increased expression of the core gene PKM is linked with a poor prognosis in a number of malignancies.