Transcriptomics Based Network Analyses and Molecular Docking Highlighted Potentially Therapeutic Biomarkers for Colon Cancer.

Abstract

In this study, machine learning-based multiple bioinformatics analysis was carried out for the purpose of the deep and efficient mining of high-throughput transcriptomics data from the TCGA database. Compared with normal colon tissue, 2469 genes were significantly differentially expressed in colon cancer tissue. Gene functional annotation and pathway analysis suggested that most DEGs were functionally related to the cell cycle and metabolism. Weighted gene co-expression network analysis revealed a significant module and the enriched genes that were closely related to fatty acid degradation and metabolism. Based on colon cancer progression, the trend analysis highlighted that several gene sets were significantly correlated with disease development. At the same time, the most specific genes were functionally related to cancer cell features such as the high performance of DNA replication and cell division. Moreover, survival analysis and target drug prediction were performed to prioritize reliable biomarkers and potential drugs. In consideration of a combination of different evidence, four genes (ACOX1, CPT2, CDC25C and PKMYT1) were suggested as novel biomarkers in colon cancer. The potential biomarkers and target drugs identified in our study may provide new ideas for colonic-related prevention, diagnosis, and treatment; therefore, our results have high clinical value for colon cancer.