miRNAFinder: A comprehensive web resource for plant Pre-microRNA classification

Area of Science:

• Bioinformatics
• Molecular Biology
• Computational Biology

Background:

• MicroRNAs (miRNAs) are small non-coding RNAs regulating gene expression at RNA or DNA levels.
• Precursor miRNAs (pre-miRNAs) possess distinct hairpin loop structures, making them targets for computational identification.
• Existing experimental methods for miRNA identification have limitations, driving the need for advanced computational approaches.

Purpose of the Study:

• To develop and present a novel multilayer perceptron (MLP) based classifier for accurate plant pre-miRNA identification.
• To introduce a new, curated dataset for training and testing machine learning models, specifically addressing data overlap issues in existing plant pre-miRNA datasets.
• To provide a freely accessible web server for the developed classifier, applicable to any plant species.

Main Methods:

• Implementation of a multilayer perceptron (MLP) classifier utilizing 180 features across sequential, structural, and thermodynamic categories.
• Development of a novel dataset to overcome positive training and testing data overlap issues present in the PlantMiRNAPred dataset.
• Validation of the MLP model on plant pre-miRNA identification and testing on other small non-coding RNA types.

Main Results:

• The developed MLP classifier achieved 92% accuracy, 94% specificity, and 90% sensitivity for plant pre-miRNA identification.
• The model demonstrated 78% accuracy when tested on other types of small non-coding RNAs.
• A new dataset was created and utilized, improving the reliability of machine learning models for real and pseudo-plant pre-miRNA classification.

Conclusions:

• The study successfully developed a highly accurate computational tool for identifying plant pre-miRNAs using an MLP classifier and a novel dataset.
• The introduced classifier and dataset offer a valuable resource for plant miRNA research, enhancing the accuracy and reliability of in-silico pre-miRNA identification.
• The freely available web server (http://mirnafinder.shyaman.me/) facilitates broader accessibility and application of this tool across diverse plant species.