miREM: an expectation-maximization approach for prioritizing miRNAs associated with gene-set

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• MicroRNAs (miRNAs) are key regulators of gene expression, influencing diverse cellular mechanisms.
• The complex regulatory network, where one miRNA targets many genes and one gene is targeted by multiple miRNAs, poses challenges for accurate prediction.
• Understanding these intricate miRNA-gene interactions is crucial for deciphering cellular processes.

Purpose of the Study:

• To introduce miREM, a computational tool designed to improve the prediction and prioritization of miRNAs from user-defined gene sets.
• To provide an accessible web-server with an intuitive graphical user interface for miRNA target prediction.

Main Methods:

• miREM couples an expectation-maximization (EM) algorithm with hypergeometric probability (HP) for enhanced prediction accuracy.
• The tool integrates a comprehensive compendium of human and mouse miRNA-target prediction databases.
• Users can upload gene sets, filter predictions (e.g., non-conserved miRNAs), and visualize results through interactive plots and heatmaps.

Main Results:

• miREM successfully predicted manipulated miRNAs in RNA sequencing datasets from knock-in and knock-out experiments.
• The tool's predictions were validated against existing miRNA prediction programs, showing comparable or superior performance.
• Results are presented via a graphical interface, enabling prioritization, visualization, and filtering of predicted miRNA-gene interactions.

Conclusions:

• miREM offers a robust and user-friendly platform for predicting and prioritizing miRNAs.
• The tool enhances the analysis of miRNA regulatory networks, contributing to a deeper understanding of gene expression.
• miREM demonstrates improved accuracy and utility compared to existing miRNA prediction tools.