Redundancy analysis allows improved detection of methylation changes in large genomic regions

Area of Science:

• Epigenetics and Genomics
• Computational Biology
• Disease Biomarker Discovery

Background:

• DNA methylation is a key epigenetic regulator of gene expression, influenced by environmental factors and implicated in disease.
• Current methods for detecting differentially methylated regions (DMRs) are limited to small regions (kilobases) and cannot assess large genomic areas.
• Genomic structural variants can impact methylation across megabase-scale regions, necessitating methods capable of analyzing larger targets.

Purpose of the Study:

• To develop a novel computational approach for identifying differential DNA methylation in large, user-defined genomic regions.
• To overcome the limitations of existing DMR detection methods that are restricted to small regions.
• To provide a robust tool for analyzing large-scale methylation changes associated with specific outcomes or exposures.

Main Methods:

• Developed a new DMR detection approach based on redundancy analysis (RDA).
• The method assesses differential methylation within targeted genomic regions of interest.
• Implemented the methodology in the MEAL Bioconductor package for methylation data analysis.

Main Results:

• The RDA-based method demonstrated superior performance compared to Bumphunter, blockFinder, and DMRcate in simulated and real datasets.
• Achieved high accuracy, controlled type I error, and performed well even with small sample sizes and subtle methylation changes.
• The method quantifies the degree of association, allows targeted region analysis, and evaluates simultaneous effects of multiple variables.

Conclusions:

• A novel multivariate approach effectively identifies differential methylation in large genomic regions.
• The proposed method outperforms existing popular DMR detection tools in accuracy and scope.
• Capable of analyzing complex effects, including multiple variables, offering advanced insights into epigenetic regulation and disease.