Machine Learning Reveals the Association Between Gene Expression and Immune Infiltration in Colorectal Cancer: A Comprehensive Study

Abstract

Background:

Colorectal cancer (CRC) is a leading cause of cancer mortality globally. The molecular mechanisms of CRC and the accumulating immune cell infiltration in the tumor microenvironment (TME) are essential for enhancing the treatment strategy and evaluation of the prognosis. In this study, the authors applied machine learning techniques to single-cell RNA sequencing data to investigate the gene expression characteristics of immune cells in CRC and their association with immune cell infiltration.

Methods:

Differentially expressed genes (DEGs) in CRC were identified by machine learning methods, including clustering analysis, survival analysis, and gene enrichment analysis, and prognostic models were constructed. CIBERSORT and ESTIMATE algorithms were used to evaluate the abundance of infiltrating immune cells and UMAP and t-SNE techniques were used for dimensionality reduction and visualization of the data.

Results:

Specific gene expression patterns are closely related to immune cell infiltration in CRC patients. Clustering analysis demonstrated two unique subgroups in the CRC samples, characterized by significant differences in survival outcomes (p = 0.049). These DEGs are enriched in various biological processes, according to gene enrichment analysis. The prognostic models of the receiver operating characteristic curves had good predictive accuracy, with area under the curve values. Single-cell data analysis also showed the intricate associations of immune cells with tumor cells in the TME.

Conclusions:

This study reveals the complex relationship between gene expression and immune infiltration in CRC using machine learning techniques, and establishes prognostic models with potential value in the clinic. These findings reveal the new potential biomarkers for CRC desensitization and immunotherapy.