snoGloBe

snoGloBe predicts interactions between human C/D small nucleolar RNAs (snoRNAs) and target RNAs to map binding sites and potential regulatory effects including 2'-O-ribose methylation guidance and non-ribosomal gene regulation.

Key Features:

• Gradient Boosting Classifier: Uses a gradient boosting classifier to predict snoRNA–RNA interactions by analyzing target type, position, and sequence features.
• Comprehensive Interaction Analysis: Provides interaction predictions that complement large-scale RNA–RNA interaction datasets constrained by specific experimental conditions.
• Regulatory Element Enrichment: Identifies significant enrichment of predicted interactions near gene expression regulatory elements such as splice sites.
• Functional Validation and Novel Predictions: Produces predictions consistent with experimentally validated binding sites and extends to novel sites with shared regulatory functions.
• Impact on Gene Expression: Links predicted interactions to functional effects supported by experimental evidence of changes in target abundance and splicing after snoRNA knockdown.
• Overlap with RNA-Binding Proteins: Reveals frequent coincidence between predicted snoRNA interactions and binding sites of functionally related RNA-binding proteins.
• Application to Viral RNAs: Detects viral RNA targets, including snoRNA interactions with heavily methylated SARS-CoV-2 RNA.

Scientific Applications:

• Interaction mapping: Mapping snoRNA–RNA binding sites and interaction networks across transcripts.
• Regulatory mechanism studies: Investigating snoRNA roles in gene expression regulation, including effects on RNA abundance and splicing.
• RBP interplay analysis: Exploring overlaps between snoRNA interaction sites and RNA-binding protein binding sites.
• Viral-host interaction analysis: Identifying snoRNA interactions with viral RNAs such as SARS-CoV-2.
• Therapeutic target discovery: Prioritizing snoRNA–RNA interactions that may be involved in disease mechanisms.

Methodology:

Predicts snoRNA–RNA interactions using a gradient boosting classifier that integrates features of target type, target position, and sequence.