DirectRM: integrated detection of landscape and crosstalk between multiple RNA modifications using direct RNA...

Area of Science:

• Molecular Biology
• Genomics
• Bioinformatics

Background:

• RNA modifications play crucial roles in gene regulation.
• Accurate profiling of multiple RNA modifications is challenging.
• Nanopore direct RNA sequencing offers a promising avenue for native RNA analysis.

Purpose of the Study:

• To develop a robust computational tool for simultaneous detection of multiple RNA modifications.
• To leverage nanopore direct RNA sequencing for high-throughput epitranscriptomic profiling.
• To investigate the co-occurrence and spatial relationships of RNA modifications.

Main Methods:

• Developed DirectRM, a two-stage computational pipeline.
• Utilized a binary classifier to identify candidate modified k-mers.
• Employed an attention-based neural network for precise modification and position determination.
• Trained and validated the model on native RNA samples from human cell lines and viral RNAs.

Main Results:

• DirectRM accurately detects six abundant RNA modifications simultaneously in native RNAs.
• The tool demonstrates high sensitivity, precision, and robustness, surpassing existing methods.
• Revealed transcript-level proximity and molecule-level interference of adjacent RNA modifications.
• Established associations between modifications at both transcript and molecule levels.

Conclusions:

• DirectRM provides a powerful approach for comprehensive epitranscriptome analysis.
• The findings offer new insights into the complex interplay of RNA modifications.
• DirectRM is expandable for detecting additional RNA modifications in future studies.