A Supervised Ensemble Approach for Sensitive microRNA Target Prediction.

Area of Science:

• Bioinformatics
• Computational Biology
• Molecular Biology

Background:

• MicroRNAs (miRNAs) are small non-coding RNAs regulating gene expression post-transcriptionally.
• Accurate experimental methods for determining miRNA-mRNA interactions remain a challenge.
• In silico approaches are crucial for understanding miRNA target recognition.

Purpose of the Study:

• To develop a sensitive computational model for predicting miRNA target interactions.
• To explore and incorporate key miRNA target recognition features into predictive modeling.
• To create a robust tool for miRNA target prediction.

Main Methods:

• Employed a supervised ensemble under-sampling technique to handle imbalanced datasets.
• Utilized feature selection methods to identify optimal features for miRNA-mRNA target recognition.
• Developed miRTPred, a blending ensemble model using weighted voting classifiers.

Main Results:

• Achieved a sensitivity of 87% and an F1-score of 0.88 for the 3' untranslated region (3'UTR) of mRNA.
• The miRTPred model demonstrated superior performance compared to existing machine learning and non-machine learning methods.
• Identified key features influencing miRNA-mRNA target recognition.

Conclusions:

• miRTPred offers a sensitive and accurate method for predicting miRNA-mRNA interactions.
• The developed model advances the field of miRNA research by providing a reliable in silico tool.
• miRTPred is publicly available, facilitating further research in gene regulation.