Identification of microRNA Functional Targets Based on microRNA and mRNA Co-Expression Network Analysis
thesisposted on 24.02.2014, 00:00 by Bhavisha P. Chapatwala
Background: MicroRNAs are essential key regulators of gene expression. They have significance in essential biological process. MicroRNA expression patterns are promising biomarkers for several tumor types including breast cancer. Many computational approaches are proposed to classify miRNA functions in recent years. Here, we propose an integrative approach to identify miRNA modules and its functional targets through the analysis of global miRNA and mRNA expression data. Our interest is to identify functionally correlated miRNA-mRNA modules that are involved in specific biological processes. Results: The Weighted Gene Co-expression Network Analysis (WGCNA) methodology was applied to analyze miRNA and mRNA expression data in order to determine the statistically significant modules of miRNA and the function of their targets. The process can be divided into three categories: (1) identify which mRNAs were targeted by which miRNAs, (2) determination of miRNA regulatory modules, i.e. to identify a group of co-expressed miRNAs and mRNAs. (3) Investigation of the miRNA regulatory modules i.e. to find an involvement in specific biological process for a particular miRNA module. Conclusion: We used mRNA and miRNA expression data from Espen Enerly breast cancer study. The proposed framework effectively captured miRNA modules. Through Gene Ontology analysis, several biological processes involving miRNAs and their targeted mRNAs were identified. To determine coherent miRNA-mRNA modules, we demonstrated that mRNAs in one module exhibit higher correlation with the miRNAs in a module. However, due to the fact that only the small numbers of mRNA modules were detected from the WGCNA analysis for this datasets, we were not able to find other miRNA-mRNA modules. For that reason we converted our focus to the other miRNAs which are not related to any modules. Therefore, the effectiveness of this approach has to be further investigated using other datasets.