MethylAction: detecting differentially methylated regions that distinguish biological subtypes

Area of Science:

• Genomics
• Epigenetics
• Bioinformatics

Background:

• DNA methylation differences are crucial for understanding molecular and gene-regulatory states in biological subtypes.
• Enrichment-based next-generation sequencing methods like MBD-isolated genome sequencing (MiGS) and MeDIP-seq enable genome-wide DNA methylation studies.
• Existing analytical tools are limited for analyzing three or more groups, hindering comprehensive subtype comparison.

Purpose of the Study:

• To introduce MethylAction, a novel analytical tool designed for detecting DNA methylation patterns in multi-group comparisons.
• To address the limitations of current tools in analyzing three-group or larger study designs for DNA methylation data.
• To provide a robust method for identifying statistically significant differentially methylated regions (DMRs) across multiple biological subtypes.

Main Methods:

• MethylAction detects all possible patterns of statistically significant hyper- and hypo-methylation across any number of groups.
• Significance of methylation changes is determined at the level of differentially methylated regions (DMRs).
• Bootstrapping is employed to calculate false discovery rates (FDRs) for each identified methylation pattern.

Main Results:

• MethylAction was demonstrated in a four-group comparison of benign prostate and three prostate cancer subtypes.
• The tool identified robust patterns of DMRs, with bootstrap FDRs proving useful for selection.
• Compared to two-group comparison tools, MethylAction detected more DMRs with strong differential methylation, validated by whole genome bisulfite sequencing.
• MethylAction demonstrated a superior balance between precision and recall in cross-cohort comparisons.

Conclusions:

• MethylAction effectively addresses the need for analyzing DNA methylation in studies with three or more groups.
• The tool enhances the detection and validation of differentially methylated regions, improving the distinction between biological subtypes.
• MethylAction offers improved precision and recall for cross-cohort comparisons, advancing the analysis of epigenomic data.