Epitranscriptomic subtyping, visualization, and denoising by global motif visualization.

Area of Science:

• Molecular Biology
• Bioinformatics
• Genomics

Background:

• High-throughput sequencing enables single-base resolution epitranscriptomic profiling.
• Distinguishing true RNA modification signals from noise and subtypes is challenging with current tools.
• Existing methods struggle with visualization, denoising, and subtype identification of RNA modifications.

Purpose of the Study:

• To introduce iMVP, an interactive framework for epitranscriptomic data analysis.
• To provide tools for subtyping, visualization, and denoising of RNA modification signals.
• To enable comprehensive comparison of different epitranscriptomic profiling approaches.

Main Methods:

• Development of iMVP, an interactive framework utilizing nonlinear dimension reduction and density-based partition.
• Application of iMVP to analyze mRNA m5C and ModTect variant data.
• Utilizing iMVP for comparative analysis of 8 profiling approaches for putative m6A/m6Am sites.
• Demonstration of iMVP's capability in analyzing large-scale human A-to-I editing datasets.

Main Results:

• iMVP successfully identified novel RNA modification motifs and writers.
• The framework effectively discovered false positives undetectable by traditional methods.
• iMVP facilitated a comprehensive comparison of multiple epitranscriptomic profiling approaches.
• Analysis revealed differences and patterns between true positives and artifacts in various methods.
• iMVP demonstrated scalability by analyzing a previously unmanageable large human A-to-I editing dataset.

Conclusions:

• iMVP offers a robust framework for the visualization and interpretation of epitranscriptomic data.
• The tool enhances the ability to identify true RNA modification sites and understand profiling method performance.
• iMVP advances the analysis of complex epitranscriptomic datasets, including large-scale RNA editing data.