miRNAFinder

miRNAFinder classifies plant pre-microRNAs (pre-miRNAs) using a multilayer perceptron (MLP) classifier to distinguish real versus pseudo pre-miRNAs for studies of miRNA biogenesis and regulation.

Key Features:

• MLP classifier: Employs a multilayer perceptron (MLP) based classifier for plant pre-miRNA identification.
• Feature set: Utilizes 180 features categorized into sequential, structural, and thermodynamic aspects.
• Performance metrics: Reported accuracy 92%, specificity 94%, and sensitivity 90% for plant pre-miRNA classification.
• Cross-RNA testing: Evaluated on other small non-coding RNA types with an accuracy of 78%.
• Novel dataset: Provides a dataset developed to train and test machine learning models and to address overlapping positive training and testing data previously encountered with PlantMiRNAPred.
• Species applicability: Designed to be applicable across various plant species.

Scientific Applications:

• Plant pre-miRNA identification: Discrimination of real versus pseudo plant pre-miRNAs for miRNA discovery and annotation.
• miRNA biogenesis and regulation studies: Support for research into miRNA function, maturation from pre-miRNAs, and regulatory roles in plants.
• Machine learning development: Use of the novel dataset for training, testing, and benchmarking ML classifiers for RNA classification.
• Small non-coding RNA classification: Application to classify other small non-coding RNAs, as demonstrated by reported cross-RNA accuracy.

Methodology:

Uses a multilayer perceptron (MLP) classifier trained and evaluated on a novel dataset, employing 180 features categorized as sequential, structural, and thermodynamic.