rG4detector

rG4detector detects potential RNA G-quadruplexes (rG4s) in transcriptomics data using a convolutional neural network to predict their presence and stability for analysis of rG4 roles in RNA biology.

Key Features:

• Convolutional Neural Network Architecture: Utilizes a CNN framework to analyze transcriptomics data and identify potential rG4-forming sequences.
• Enhanced Prediction Accuracy: Demonstrates improved performance over existing methods in predicting both the presence and stability of rG4s.
• Integration of Stability Information: Incorporates rG4 stability information into the predictive model to address limitations of sequence-only approaches.

Scientific Applications:

• G3BP1–rG4 interaction analysis: Applied to study interactions between rG4s and the RNA-binding protein G3BP1.
• Dynamic enrichment analysis: Revealed dynamic enrichment of rG4s bound by G3BP1 during cellular stress and in stress granules (SGs).
• Experimental validation and functional insight: Supported experimental findings that G3BP1 is an rG4-binding protein, that endogenous rG4s are enriched within SGs, and that reduced rG4 availability impairs SG formation.

Methodology:

Uses a convolutional neural network to process and analyze large-scale transcriptomics data and integrates sequence features with rG4 stability information.