ACGLM: A Hybrid Approach to Select and Combine Gene Expression Regulation in Cancer Datasets

Abstract

Abstract Cancer is one of the causes of death in the world and many genes are involved in it. Transcription factors (TFs) and microRNAs (miRNAs) are primary gene regulators and regulatory mechanisms for cells to define their targets. The study of the Regulatory mechanisms of the two main regulators is complex, but this lead to a deeper interpretation of biological processes. In order to avoid exhaustive search and unnecessary genes, firstly, mRNA expression and miRNA expression are clustered by K-means cluster, then, applied ANOVA test to select significant genes. We proposed a gene regulatory network (GRN) estimation method, using Directed networks with generalized linear regression to predict and explain the relationships between regulators and their targets. Where through GO TERM and KEGG PATHWAY for target genes we got many processes such as cell communication, regulation of the biologic process, biological regulation and cell cycle, DNA replication, and cell cycle, these processes are considered significant to the cancer diseases. by comparing with other methodologies Our approach was better, as well as the results were consistent with the medical literature, where the important regulators in our gene regulatory network have a major role in cancer this explains the efficiency of this approach.