DeepRNA-Reg: a deep-learning based approach for comparative analysis of CLIP experiments.

Area of Science:

• Computational Biology
• Genomics
• Molecular Biology

Background:

• High-throughput sequencing of RNA isolated by crosslinking immunoprecipitation (HITS-CLIP) is crucial for studying RNA-protein interactions.
• Analyzing differential HITS-CLIP data, especially when microRNA (miRNA) activity is modulated, presents analytical challenges.
• Understanding miRNA-mediated RNA targeting requires accurate prediction of RNA structural motifs and regulatory networks.

Purpose of the Study:

• To introduce DeepRNA-Reg, a novel deep learning framework for high-fidelity comparative analysis of paired HITS-CLIP datasets.
• To evaluate DeepRNA-Reg's performance against existing methods for differential HITS-CLIP analysis.
• To identify novel mediators of miRNA-mediated regulation in biological systems using DeepRNA-Reg.

Main Methods:

• Development of DeepRNA-Reg, a deep learning model leveraging advances in AI for HITS-CLIP data analysis.
• Application of DeepRNA-Reg to paired HITS-CLIP datasets with perturbed miRNA activity (e.g., gene knockout of miRNA clusters).
• Comparative analysis of DeepRNA-Reg's predictions against established differential HITS-CLIP analysis methods and ground-truth RNA structural data.

Main Results:

• DeepRNA-Reg demonstrated superior prediction accuracy compared to current leading methods for differential HITS-CLIP analysis.
• Predictions generated by DeepRNA-Reg showed better adherence to known RNA primary and secondary structural motifs involved in miRNA targeting.
• The tool successfully uncovered novel mediators in the mechanism of miRNA-mediated restraint of type-2 immunity in T-Helper 2 cells.
• DeepRNA-Reg predictions exhibited enhanced translatability across different biological contexts.

Conclusions:

• DeepRNA-Reg offers a robust and accurate deep learning approach for analyzing HITS-CLIP data, particularly in comparative studies.
• The framework improves the understanding of miRNA-mediated RNA regulation by accurately predicting structural motifs and identifying novel regulatory elements.
• DeepRNA-Reg provides a versatile tool with broad applicability for researchers investigating RNA biology and gene regulation across diverse biological systems.