Augmented prediction of multi-species protein-RNA interactions using evolutionary conservation of RNA-binding...

Area of Science:

• Genomics
• Computational Biology
• Molecular Biology

Background:

• RNA-binding proteins (RBPs) are crucial regulators of gene expression.
• Experimental profiling of RBP-RNA interactions is challenging across many species.
• Understanding cross-species RBP binding is vital for comparative genomics and disease research.

Purpose of the Study:

• To develop a computational framework for predicting RNA-binding protein (RBP)-RNA interactions across diverse species.
• To leverage evolutionary conservation for enhanced prediction accuracy.
• To quantify the impact of genetic variants on RBP binding.

Main Methods:

• Introduced MuSIC (Multi-Species RBP-RNA Interactions using Conservation), a deep learning framework.
• Utilized label smoothing and evolutionary conservation data from 11 species (human to yeast).
• Validated predictions through experimental methods and pathway enrichment analysis.

Main Results:

• MuSIC significantly outperforms existing computational methods in predicting RBP-binding peaks across species.
• Prediction confidence is higher in metazoan species, correlating with RBP conservation patterns.
• Quantified homologous genetic variant effects on RBP binding, identifying disrupted binding in ubiquitination pathways.

Conclusions:

• MuSIC offers a robust computational approach for predicting cross-species RBP-RNA interactions.
• The framework provides insights into RBP-mediated gene regulation and its role in human diseases.
• Enables computational quantification of genetic variant effects on RBP binding across species.