GraPPLE

GraPPLE classifies non-coding RNA (ncRNA) sequences by extracting graph-based features from predicted RNA secondary structures and applying Support Vector Machines (SVMs) to predict functionality and assign RNAs to one of 46 Rfam families.

Key Features:

• Graph Representation: Represents predicted RNA secondary structures as graphs where nodes correspond to nucleotides or structural elements and edges denote interactions.
• Functional Classification: Classifies ncRNA sequences as functional or non-functional and assigns functional RNAs to one of 46 Rfam families.
• Support Vector Machines (SVMs): Uses Support Vector Machines (SVMs) trained on graph-derived features to perform classification.
• Robustness Against Sequence Divergence: Demonstrates enhanced robustness to sequence divergence compared with sequence-similarity-based methods and covariance models.
• Integration with Existing Methods: Can be combined with existing prediction methodologies to improve overall accuracy.
• Insight into Structural Features: Identifies informative graph properties that provide insight into structural determinants of RNA functionality.

Scientific Applications:

• Candidate filtering: Filters potentially interesting ncRNA candidates from large genomic or sequencing datasets based on predicted structural features.
• Functional annotation: Supports annotation of non-coding genomic regions by predicting RNA functionality and Rfam family membership.
• RNA biology research: Aids studies of genomic transcription and RNA biology by characterizing structural determinants of RNA function.
• Complementary prediction: Improves reliability of ncRNA prediction for divergent sequences when used alongside sequence-based or covariance-model approaches.

Methodology:

Predicted RNA secondary structures are converted into graphs (nodes = nucleotides or structural elements; edges = interactions), graph properties are computed, and Support Vector Machines (SVMs) are trained on those features to classify sequences as functional or non-functional and to assign functional RNAs to one of 46 Rfam families.