Discovery of microRNAs and Transcription Factors Co-Regulatory Modules by Integrating Multiple Types of Genomic Data.

It is well known that regulators known as microRNA (miRNA) and transcription factor (TF) have been found to play an important role in gene regulation. However, there are few researches of collaborative regulatory (co-regulatory) mechanism between miRNA and TF on system level (function level). Meanwhile, recent advances in high-throughput genomic technologies have enabled researchers to collect diverse large-scale genomic data, which can be used to study the co-regulatory mechanism between miRNA and TF. In this paper, we propose a novel method called Sparse Network regularized non-negative matrix factorization for co-regulatory modules identification which adopts multiple non-negative matrix factorization framework to identify co-regulatory modules including miRNAs, TFs and genes. This method jointly integrates miRNA, TF and gene expression profiles, and additional priori networks were added in a regularized manner. In addition, to avoid the sparsity of these networks, we employ the sparsity penalties to the variables to achieve modular solutions. The mathematical formulation can be effectively solved by an iterative multiplicative updating algorithm. We apply this method to multiple genomic data including the expression profiles of miRNAs, TFs and genes on breast cancer obtained from TCGA, priori miRNA-gene regulations, TF-gene regulations and gene-gene interactions. The results show that the miRNAs, TFs and genes of the co-regulatory modules are significantly associated and modules have a reasonable size distribution. Furthermore, the co-regulatory modules are significantly enriched in Gene Ontology biological processes and Kyoto Encyclopedia of Genes and Genomes pathways, respectively.