Biclustering analysis of transcriptome big data identifies condition-specific microRNA targets

Area of Science:

• Genomics and Bioinformatics
• Molecular Biology
• Cancer Research

Background:

• MicroRNAs (miRNAs) are crucial regulators of gene expression.
• Identifying miRNA regulatory modules (targets and cellular contexts) is essential for understanding biological processes and diseases.
• Existing methods for identifying miRNA targets have limitations in accuracy and scope.

Purpose of the Study:

• To present a novel biclustering approach for identifying human miRNA regulatory modules.
• To enhance the accuracy of miRNA target prediction by integrating mRNA fold-change data and functional networks.
• To analyze cancer-specific miRNA regulatory modules and identify potential therapeutic targets.

Main Methods:

• Biclustering of large-scale mRNA fold-change data for sequence-specific miRNA targets.
• Assessment of bicluster targets using validated messenger RNA (mRNA) targets.
• Incorporation of functional networks to refine target identification.
• Analysis of cancer-specific biclusters, including enrichment analysis of signaling pathways.
• Experimental validation of identified miRNA-target interactions and pathway modulation.

Main Results:

• The novel biclustering approach achieved an average of 17.0% improved certainty (sensitivity + specificity) in identifying miRNA targets.
• Incorporating functional networks further increased the net gain in certainty to an average of 32.0%.
• Analysis revealed enrichment of the PI3K/Akt signaling pathway in breast cancer and diffuse large B-cell lymphoma, regulated by specific miRNAs.
• Five independent prognostic miRNAs were identified, and the repressive effect of miR-29 on bicluster targets and pathway activity was experimentally validated.
• The BiMIR database was created, containing 29,898 biclusters for 459 human miRNAs, searchable by various parameters.

Conclusions:

• The developed biclustering method significantly enhances the accuracy and scope of human miRNA regulatory module identification.
• This approach facilitates the discovery of novel miRNA-target interactions and their roles in complex diseases like cancer.
• The findings highlight the importance of the PI3K/Akt pathway in specific cancers and identify prognostic miRNAs, offering potential for therapeutic strategies.