Heterogeneous ensemble approach with discriminative features and modified-SMOTEbagging for pre-miRNA classification

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• MicroRNA precursors (pre-miRNAs) play crucial roles in gene regulation.
• Accurate classification of pre-miRNAs is essential for understanding their function.
• Existing methods often struggle with species-specific variations and feature selection.

Purpose of the Study:

• To develop a robust and accurate ensemble classifier for pre-miRNA identification.
• To improve classification performance using discriminative structural features.
• To validate the model's effectiveness across different species.

Main Methods:

• An ensemble classifier combining Support Vector Machine (SVM), k-nearest neighbors (kNN), and Random Forest (RF) algorithms.
• Utilizing discriminative features, self-containment, and derivatives for structural analysis.
• Employing Correlation-based Feature Selection (CFS) with Genetic Algorithm (GA) and a modified Synthetic Minority Oversampling Technique (SMOTE) bagging for preprocessing.

Main Results:

• Achieved an overall prediction accuracy of 96.54% with 10 runs of 5-fold cross-validation.
• Demonstrated high sensitivity (94.8%) and specificity (98.3%), outperforming state-of-the-art methods.
• Showed high accuracy (>93%) when applied to animal, plant, and virus pre-miRNAs.

Conclusions:

• The heterogeneous ensemble classifier provides more reliable pre-miRNA predictions than single classifiers.
• Selected features indicate high intrinsic structural robustness of pre-miRNAs.
• The developed model offers a significant advancement in pre-miRNA classification across diverse species.