CRMSS

CRMSS predicts binding sites between circular RNAs (circRNAs) and RNA-binding proteins (RBPs) to identify molecular interaction sites relevant to disease mechanisms.

Key Features:

• Multi-Scale Characterization: Integrates multi-scale sequence and structural characterization of circRNAs and RBPs to improve binding site prediction precision.
• Sequence Embedding for circRNAs: Uses sequence k-mer embedding and assesses forming probabilities of local secondary structures to represent circRNA features.
• Feature Generation for RBPs: Combines sequence and structure frequencies derived from RNA-binding domain regions to represent RBP characteristics.
• Advanced Pattern Recognition: Employs multi-scale residual blocks to capture intricate circRNA–RBP binding patterns.
• Contextual Information Extraction: Applies BiLSTM (Bidirectional Long Short-Term Memory) networks and attention mechanisms to extract high-level contextual information.
• Validation and Performance: Validated against 37 different RBPs, demonstrating superior performance to existing state-of-the-art methods and accurately identifying experimentally verified circRNA–RBP pairs.

Scientific Applications:

• Molecular mechanism studies: Enables investigation of circRNA–RBP interactions implicated in disease-related molecular mechanisms.
• Therapeutic target identification: Supports identification of potential therapeutic targets by predicting circRNA–RBP binding sites relevant to disease pathology.

Methodology:

Analyzing circRNAs via sequence k-mer embedding and local secondary structure forming probabilities; characterizing RBPs by integrating sequence and structure frequencies from RNA-binding domain regions; utilizing multi-scale residual blocks; and applying BiLSTM networks with attention mechanisms for contextual analysis.