m⁶Aexpress-enet: Predicting the regulatory expression m⁶A sites by an enet-regularization negative binomial...

Area of Science:

• Molecular Biology
• Genomics
• Bioinformatics

Background:

• N6-methyladenosine (m6A) is the most abundant mRNA modification, influencing mRNA metabolism, stability, and degradation.
• The precise role of individual m6A sites in gene expression regulation remains largely unknown.
• Predictive modeling is challenged by multicollinearity arising from correlated methylation levels across multiple m6A sites within a gene.

Purpose of the Study:

• To develop a robust statistical model for predicting the regulatory potential of specific m6A sites on gene expression.
• To address the multicollinearity issue inherent in analyzing multiple m6A sites per gene.
• To provide a tool for uncovering complex m6A regulatory patterns and mechanisms.

Main Methods:

• Proposed an elastic-net regularized negative binomial regression model, termed m6Aexpress-enet.
• Evaluated model performance using simulated datasets to demonstrate predictive accuracy.
• Applied m6Aexpress-enet to real MeRIP-seq data from human lymphoblastoid cell lines.

Main Results:

• m6Aexpress-enet achieved top prediction performance on simulated data.
• Analysis of real data revealed complex regulatory patterns of predicted m6A sites.
• Identified unique enrichment pathways within constructed co-methylation modules.

Conclusions:

• m6Aexpress-enet is a powerful tool for predicting m6A site-specific gene expression regulation.
• The model facilitates the discovery of mechanisms underlying m6A-mediated gene expression control.
• Source code and implementation details are publicly available for broader scientific use.