NCodR

NCodR classifies non-coding RNAs in Viridiplantae using a multi-class Support Vector Machine that leverages single sequence repeat signatures to distinguish ncRNA types.

Key Features:

• Multi-class SVM (RBF): Uses a multi-class Support Vector Machine with a radial basis function kernel and achieves an average F1 score of approximately 91%.
• Training features: Trained on attributes derived from sequence and structural analyses.
• AU content distribution: Identifies distinct AU content regions for different ncRNA classes, with some overlap between classes.
• Minimum Folding Energy Index (MFEI): Reports similar average MFEI values across most ncRNA classes while pre-miRNAs and long non-coding RNAs (lncRNAs) exhibit distinct MFEI patterns.
• k-mer repeat signatures (3-mer): Analyzes k-mer repeat signatures of length three to reveal unique k-mer distribution patterns among ncRNA types.
• Single sequence repeat signatures: Leverages single sequence repeat signatures as differentiating features for ncRNA classification.
• Classification scope: Distinguishes seven distinct ncRNA classes within Viridiplantae.

Scientific Applications:

• Gene regulation studies: Enables analysis of ncRNA roles in gene expression regulation in plants by accurately classifying ncRNA types.
• Functional annotation: Supports functional annotation of plant transcriptomes by assigning ncRNA sequences to specific classes.
• Structure–function investigations: Facilitates computational structural biology and molecular biology studies into ncRNA structure–function relationships and interactions.

Methodology:

Multi-class Support Vector Machine with a radial basis function kernel trained on sequence- and structure-derived features including AU content distribution, Minimum Folding Energy Index (MFEI), and 3-mer repeat (single sequence repeat) signatures, yielding an average F1 ≈ 91%.