Feature Selection and mRNA-miRNA bipartite Network Analysis for Biomarker Development and Drug Repurposing in Colon Cancer

Abstract

Background: Colon cancer (CC) is the second cause of mortality among other non-skin cancers. Despite the progression in screening and diagnostic strategies, further information is still demanding for diagnosing, managing, and treating CC. In order to achieve this goal, computational analysis, including data mining algorithms are essential tools and the most widely used approach for achieving important features of gene expression data combined with clinical assessments which paves the way for diagnosis and treatment at the right time. Methods: This study aimed to identify the most critical genes and candidate drugs that may be helpful for CC treatment and develop novel microRNA biomarkers for CC. The gene expression data from CC tissues, including 98 CC and 98 normal samples, were used to reconstruct gene co-expression networks. Deferentially expressed genes (DEGs) were extracted after preprocessing and normalization. Then, DEGs were used to construct the weighted gene co-expression network (WGCNA), and low-preserved modules were obtained. The experimentally validated mRNA-miRNA interaction information was used to construct three bipartite mRNA-miRNA networks. From each bipartite network, ten hub miRNAs were extracted for further analysis. Finally, a drug-gene interaction network was constructed. Results: The most significant genes were ADH1B, AQP8, CA1, CLCA4, GUCA2B, MMP7, MS4A12, and TMIGD1 obtained from feature selection. The data included novel miRNAs (e.g., hsa-miR-665, hsa-miR-30c-3p, hsa-let-7b-5p, hsa-miR-3689d, hsa-miR-1233-5p, hsa-miR-1910-3p, and finally hsa-miR-6788-5p which was thoroughly novel without any investigation in other cancers) and might be potentially associated with CC. Also, the most important genes were targets for these miRNAs. Eventually, drugs such as Cytarabine, Levodopa, Methyldopa, Tretinoin, Hydralazine, Daunorubicin Hydrochloride, Olanzapine, and Idarubicin were proposed as potential novel drugs for CC treatment. Conclusion: The present study was performed to determine the most significant genes, develop biomarkers, and find novel potential drugs for colon cancer. All potential biomarkers and drugs obtained from this study could be practical after extensive experimental studies for using them in clinical experiments.