A Two-Stage Method to Identify Joint Modules From Matched MicroRNA and mRNA Expression Data

MicroRNAs (miRNAs) are a class of small non-coding RNAs, which play key roles in gene regulation. Previous studies have revealed that they are likely to work together to regulate their common target genes. Discovery of the underlying combinatorial regulation between miRNAs and genes can provide useful information for understanding their functions as well as their close relevance to cancer. In this paper, we propose a two-stage method for identifying miRNA-gene regulatory modules by integrating miRNA and mRNA expression profiles and miRNA genomic cluster data. We first develop a multiple-output sparse group lasso (MSGL) regression model to predict a miRNA-gene association matrix (i.e., a miRNA-gene regulatory network). Further, we propose a L0-regularized singular value decomposition (L0-SVD) to identify miRNA-gene joint modules from the predicted regulatory network. We test our method on the matched miRNA and mRNA expression profiles in breast cancer from TCGA project and identify ten miRNA-gene regulatory modules. We find that 1) the modules are significantly associated in the predicted miRNA-gene regulatory network; 2) the modules are significantly enriched in GO biological processes and KEGG pathways, respectively; 3) many miRNAs and genes in the modules are related with breast cancer. On average, 51% of the miRNAs and 30% of the genes are related with breast cancer. The results demonstrate that miRNA-gene regulatory modules provide insights into the mechanisms of the combinatorial regulation between miRNAs and genes.