Bayesian inference of MicroRNA targets from sequence and expression data.

Area of Science:

• Genomics
• Molecular Biology
• Bioinformatics

Background:

• MicroRNAs (miRNAs) are crucial regulators of mammalian gene expression, targeting messenger RNAs (mRNAs) to control protein translation.
• Computational prediction methods identify numerous potential miRNA targets, but distinguishing functional targets remains a challenge.

Purpose of the Study:

• To develop a novel Bayesian model, GenMiR++, for efficiently identifying high-confidence functional miRNA targets from computational predictions.
• To improve the understanding of miRNA-mediated gene regulation.

Main Methods:

• Developed GenMiR++, a generative Bayesian model integrating miRNA expression data and candidate target information.
• Employed a Bayesian learning algorithm to score and rank potential miRNA targets.
• Validated the model using mouse data and known miRNA-target interactions.

Main Results:

• GenMiR++ identified 467 high-confidence miRNA targets from 1,770 candidates at a 2.5% false detection rate.
• Confirmed interactions include miR-92 targeting MAP2K4 and miR-16 targeting BCL2.
• The model demonstrated robustness against data perturbations.

Conclusions:

• GenMiR++ effectively identifies functional miRNA targets, significantly increasing the number of validated interactions.
• Provides a valuable starting point for a comprehensive understanding of miRNA-driven gene regulation.