An in silico analysis of microRNAs: mining the miRNAome

Murray BS, Choe SE, Woods M, Ryan TE, Liu W.

Mol Biosyst. 2010 Oct; 6(10):1853-62

PMID: 20539892

http://www.ncbi.nlm.nih.gov/pubmed/20539892

Abstract

Systematic analysis of literature- and experimentally-derived datasets using text mining with ontological enrichment and network modeling revealed global trends in the microRNA (miRNA) interactome.

A total of 756 unique miRNAs were resolved from PubMed abstracts and 1165 direct relationships between 270 miRNAs and 581 genes were identified as phrase groups using semantic search techniques.

These miRNA:gene interactions were built into a bipartite network (the miRNAome) which displays scale-free degree distribution. Functional classification of miRNA-target genes using PANTHER revealed 189 distinct molecular functions, with significant enrichment of nucleic acid binding, transcription and protein phosphorylation. Pathway analysis revealed a network of 176 miRNAs linked to 368 OMIM disorders via their target genes, which are enriched (p = 0.0047) for disease-associated SNP variations.

Reference to a database of drug targets revealed that 24.8% of all published miRNA-targets are targets for drug development programs, while a sub-set (18.2%) are targets for FDA-approved drugs. Consistent with topological analysis of the miRNA-disease network, the most prevalent class of FDA-approved drugs is anti-neoplastic agents against published miRNA-target genes. Linking miRNAs to biological process and diseases reveals distinct co-regulation of phenotypes that could aid in understanding the role miRNA-based gene regulation plays in biological phenomena.