Large-scale inference of competing endogenous RNA networks with sparse partial correlation

Area of Science:

• Genomics
• Bioinformatics
• Molecular Biology

Background:

• MicroRNAs (miRNAs) are key regulators in biological processes and diseases.
• Complex gene regulatory networks involve competing endogenous RNAs (ceRNAs).
• Existing ceRNA analysis methods lack statistical rigor and fail to account for multiple miRNA interactions.

Purpose of the Study:

• To introduce SPONGE, a novel method for constructing ceRNA networks.
• To address limitations in existing ceRNA analysis, including statistical confounders and multiple miRNA interactions.
• To identify novel ceRNAs as potential cancer biomarkers.

Main Methods:

• SPONGE utilizes 'multiple sensitivity correlation' for ceRNA network construction.
• A probabilistic model quantifies multiple miRNA contributions to ceRNA interactions.
• P-value calculation is optimized for speed and accuracy.

Main Results:

• SPONGE accurately quantifies miRNA contributions and outperforms existing methods.
• Application to The Cancer Genome Atlas data revealed novel protein-coding and non-coding ceRNAs.
• Identified ceRNAs show potential as cancer biomarkers.

Conclusions:

• SPONGE provides a robust and efficient method for ceRNA network analysis.
• The identified ceRNAs may serve as valuable biomarkers for cancer diagnosis and prognosis.
• SPONGE facilitates a deeper understanding of miRNA-mediated gene regulation in disease.