Interpretable prediction models for widespread m6A RNA modification across cell lines and tissues

Area of Science:

• Genomics and Bioinformatics
• Molecular Biology
• Computational Biology

Background:

• RNA N6-methyladenosine (m6A) modifications are crucial for various biological processes in Homo sapiens.
• Accurate identification of m6A sites is essential for understanding their functions and molecular mechanisms.
• Current computational methods for m6A site prediction have limitations in cell line coverage and model interpretability.

Purpose of the Study:

• To develop a deep learning-based computational tool for predicting single-nucleotide-resolution m6A RNA modification sites.
• To enhance the coverage of cell lines and tissues for m6A site prediction.
• To improve the interpretability of prediction models for m6A modifications.

Main Methods:

• Developed CLSM6A, a suite of deep learning models for m6A site prediction.
• Applied models to predict m6A sites across eight cell lines and three tissues.
• Conducted extensive benchmarking against state-of-the-art methods.

Main Results:

• CLSM6A demonstrated superior performance compared to existing methods.
• The tool achieved accurate prediction of m6A sites across diverse cell lines and tissues.
• CLSM6A provides interpretable insights by identifying critical motifs and important sequence positions.

Conclusions:

• CLSM6A offers a powerful and interpretable approach for predicting m6A RNA modification sites.
• The tool's enhanced portability and interpretability advance the field of RNA modification analysis.
• CLSM6A facilitates deeper understanding of m6A functions and regulatory mechanisms.