Improving Bioinformatics Prediction of microRNA Targets by Ranks Aggregation

Area of Science:

• Bioinformatics
• Molecular Biology
• Genomics

Background:

• MicroRNAs (miRNAs) are key regulators of gene expression, impacting cellular activities by downregulating target messenger RNAs (mRNAs).
• Accurate prediction of miRNA targets is crucial for understanding gene regulation but remains a significant bioinformatics challenge due to limitations in existing software, including low accuracy, sensitivity, and inconsistent results.

Purpose of the Study:

• To develop a robust method for enhancing the accuracy and reliability of microRNA target prediction.
• To create a user-friendly webtool, miRabel, that integrates predictions from multiple established algorithms to provide a consolidated and improved list of miRNA targets.

Main Methods:

• Aggregated human miRNA target prediction results from four prominent algorithms: miRanda, PITA, SVmicrO, and TargetScan.
• Developed a novel reranking strategy incorporating additional characteristics to create a unified prediction list.
• Validated the performance of miRabel using receiver operating characteristic (ROC) and precision-recall curves with experimentally validated data and large datasets.

Main Results:

• miRabel significantly improved the accuracy and sensitivity of miRNA target predictions compared to individual algorithms.
• miRabel demonstrated superior performance against other popular prediction tools, including MBSTAR, miRWalk, ExprTarget, and miRMap.
• F-score analysis indicated that miRabel enhances the relevance of top-ranked miRNA target predictions.

Conclusions:

• The aggregation of results from multiple miRNA target prediction databases is a powerful and generalizable approach to improve prediction accuracy.
• miRabel serves as an efficient tool for biologists, facilitating the identification and contextualization of miRNA targets.
• The miRabel webtool offers a valuable resource for miRNA research, supporting the integration of predictions into biological understanding.