Analysis microRNA-gene co-modules in glioblastoma multiforme based on integrative two types of genomic data

Glioblastoma multiforme (GBM) is the most common primary central nervous system tumor. Although treatment continues to make progress in the past, the pathogenesis is still unknown to a large extent. MicroRNA (miRNA) is known to regulate cooperatively the expression of gene profiles in the post transcriptional stages and affect many crucial biological processes. The majority of the existing researches on GBM transcriptional regulation mechanism devoted to global analysis only on gene expression data, disregarding the complex coordination between miRNA and gene expression profiles. With the development of high throughput biotechnology and computational biology, it is possible to simulate and understand the regulatory role in cellular process by integrating gene expression profiles with miRNA data. In this article, we adopt the joint Non-negative Matrix Factorization Algorithm (JNMF) to integrate miRNA data and gene expression data to form co-modules. In addition, the characteristic co-modules and the number of miRNA related with GBM were identified by correlation analysis, thus constructing miRNA-gene regulation network to speculate the pathogenesis of GBM. The results demonstrated that the characteristic co-modules showed a significant biological correlation and potential association with GBM. Furthermore, GO biological processes and KEGG enrichment analysis further revealed the biological functions of miRNA targets gene in co-modules.