MatureBayes

MatureBayes predicts the precise locations of mature microRNAs within precursor sequences by applying a Naive Bayes classifier that combines sequence and secondary structure features.

Key Features:

• Naive Bayes classifier: Employs a Naive Bayes classifier to predict mature miRNA start positions.
• Sequence and secondary structure integration: Integrates primary sequence and precursor secondary structure information from miRNA precursor sequences.
• Training with positive and negative examples: Trains on positive examples (true mature miRNAs) and negative examples (same-size non-mature sequences) to optimize sensitivity and specificity.
• Position-specific feature analysis: Uses the triplet of positions 7, 8, and 9 measured from the end of the mature miRNA toward the closest hairpin as discriminatory features.
• Uracil enrichment at diagnostic positions: Observes that the 7–9 triplet is relatively conserved, predominantly contains Uracil, and is typically within or adjacent to the hairpin loop.
• Functional strand prediction: Identifies the functional strand(s) of miRNA precursors.
• Improved accuracy and generalization: Reports superior performance in predicting start positions of experimentally verified mature miRNAs, including human and mouse, with generalization across other organisms.

Scientific Applications:

• Mature miRNA localization: Precise prediction of mature miRNA start positions within precursor hairpins.
• Strand selection analysis: Identification of guide/functional strand(s) in miRNA duplexes.
• Dicer recognition studies: Investigation of sequence and structural signals (including the 7–9 triplet) implicated in Dicer processing and recognition.
• Cross-species miRNA annotation: Enhanced annotation and validation of mature miRNAs in human, mouse, and other organisms.

Methodology:

MatureBayes trains a Naive Bayes classifier on positive (true mature miRNAs) and negative (same-size non-mature) examples using sequence and secondary structure features, explicitly including the triplet of positions 7–9 from the mature miRNA end toward the closest hairpin, and optimizes sensitivity and specificity.