Prediction of protein-RNA interactions from single-cell transcriptomic data.

Area of Science:

• Computational Biology
• Genomics
• Molecular Biology

Background:

• Understanding protein-RNA interactions is vital for deciphering RNA biology, but current experimental methods are limited in scope.
• Existing computational approaches often neglect cell-type specificity and gene regulatory network (GRN) contexts.

Purpose of the Study:

• To evaluate GRN inference algorithms for predicting protein-RNA interactions from single-cell transcriptomic data.
• To introduce scRAPID, a pipeline integrating GRN methods with catRAPID for enhanced RBP-RNA interaction prediction.

Main Methods:

• Assessed GRN inference algorithms on single-cell RNA sequencing data.
• Integrated selected GRN algorithms with the catRAPID physical interaction prediction tool.
• Developed the scRAPID pipeline for RBP-RNA interaction detection.

Main Results:

• scRAPID accurately predicts RBP-RNA interactions from single-cell data, outperforming or matching transcription factor-target interaction inference.
• The integration of catRAPID significantly improves prediction accuracy, especially for long noncoding RNAs.
• Identified key hub RBPs, RNAs, and potential RBP-RBP interactions based on shared RNA targets.

Conclusions:

• Single-cell transcriptomics combined with GRN inference and physical interaction prediction is a powerful approach for studying RBP-RNA interactions.
• scRAPID provides a robust computational framework for discovering cell-type-specific RBP-RNA regulatory networks.
• The developed pipeline facilitates the identification of novel regulatory roles for RBPs and RNAs in cellular processes.