Tissue-specific RNA methylation prediction from gene expression data using sparse regression models.

Area of Science:

• Molecular Biology
• Genomics
• Bioinformatics

Background:

• N6-methyladenosine (m⁶A) is a prevalent RNA modification impacting biological processes.
• Accurate identification of m⁶A sites is essential for understanding epitranscriptome functions.
• Existing methods may not fully capture tissue-specific methylation patterns.

Purpose of the Study:

• To develop a novel computational approach, Exp2RM, for predicting tissue-specific m⁶A methylation levels.
• To evaluate the predictive performance of Exp2RM across diverse human tissues.
• To explore the functional implications of predicted m⁶A sites.

Main Methods:

• Developed Exp2RM, a single-site-based elastic net modeling approach.
• Utilized gene expression data to predict tissue-specific methylation levels.
• Trained models to incorporate tissue-specific methylation distribution patterns.

Main Results:

• Exp2RM demonstrated robust predictive performance, achieving a median R-squared of 0.482 across 22 human tissues.
• Incorporating tissue-specific patterns significantly improved prediction accuracy, reaching a median R-squared of 0.728.
• Functional analysis confirmed Exp2RM's ability to identify genes involved in relevant biological processes.

Conclusions:

• Tissue-specific methylation distribution is critical for accurate m⁶A site prediction.
• Exp2RM offers a valuable tool for investigating the epitranscriptome's role in various biological contexts.
• This approach provides insights into the functional significance of RNA methylation.