Wiring miRNAs to pathways: a topological approach to integrate miRNA and mRNA expression profiles

Abstract

The production rate of gene expression data is nothing less than astounding. However, with the benefit of hindsight we can assert that, since we completely ignored the non-coding part of the transcriptome, we spent the last decade to study cell mechanisms having few data in our hands. In this scenario, microRNAs, which are key post-trascriptional regulators, deserve special attention. Given the state of knowledge about their biogenesis, mechanisms of action and the numerous experimentally validated target genes, miRNAs are also gradually appearing in the formal pathway representations such as KEGG and Reactome maps. However, the number of miRNAs annotated in pathway maps are very few and pathway analyses exploiting this new regulatory layer are still lacking. To fill these gaps, we present 'micrographite' a new pipeline to perform topological pathway analysis integrating gene and miRNA expression profiles. Here, micrographite is used to study and dissect the epithelial ovarian cancer gene and miRNA transcriptome defining and validating a new regulatory circuit related to ovarian cancer histotype specificity.

Figure 1.

Graphical outline of the computational approach.

Figure 2.

Mucinous ovarian cancer circuit and qRT-PCR validations. (A) The path with the maximum score: the mucinous path. As described in the legend, miRNA–target validated interactions can be achieved from both manual curation (literature mining) and public databases. Predicted interactions are obtained from Targetscan predictions filtered by expression correlations. (B) qRT-PCR boxplot for specific genes and miRNA of the best path. Mucinous (M), clear cell (C), serous (S), endometrioid (E): ***≤0.005, **≤0.05, *≤0.1.