Identification of Candidate Genes for Colorectal Cancer Using Machine Learning and Bioinformatics Approaches

Colorectal cancer (CRC) stands as the second most prevalent cause of cancer deaths, and its incidence is rising over time. Identifying key genes is essential for diagnosing and developing effective therapeutic strategies for cancer. Several studies were conducted to determine the candidate genes of CRC but this is still not sufficient and further research is needed in this area. Thus, we aimed to identify the key candidate genes of CRC using The Cancer Genome Atlas (TCGA) dataset, applying the Kruskal-Wallis test and Bonferroni correction machine learning techniques. We successfully identified 9 candidate genes, including CALB2, GRP, KRAS, MLH1, NPR3, PPFIA4, SOX11, STAC2, and TRPA1, from 20518 genes of CRC using our model. In addition, we used bioinformatics frameworks to identify signaling pathways, gene ontological pathways, and PPI networks that reflect the functions of these candidate genes. We found 9 significant signaling pathways, 12 ontological pathways, and 3 hub genes for CRC. The diagnostic effectiveness of the candidate genes was evaluated through the receiver operating characteristic (ROC) analysis, and all candidate genes showed good performance according to area under the curve (AUC) values. Notably, the gene MLH1 demonstrated the highest AUC of .91. In fine, the findings of this study may play a role in disease management and offer a foundation for further laboratory investigations to uncover potential therapeutic targets for CRC treatment.