Estimating genome-wide DNA methylation heterogeneity with methylation patterns

Area of Science:

• Epigenetics
• Computational Biology
• Genomics

Background:

• Cellular diversity is key to understanding variable responses to the environment.
• DNA methylation patterns reflect this heterogeneity and can serve as disease biomarkers.
• Current methods for assessing cell-to-cell methylation heterogeneity have limitations.

Purpose of the Study:

• To develop and evaluate model-based methods for estimating genome-wide DNA methylation heterogeneity.
• To assess the performance of these new methods compared to existing approaches.
• To explore the utility of methylation heterogeneity as a biological marker.

Main Methods:

• Adapted a mathematical framework from biodiversity for DNA methylation analysis.
• Developed three model-based methods (MeH) for estimating methylation heterogeneity.
• Evaluated methods using synthetic and real datasets, including human cancer and Arabidopsis data.

Main Results:

• The proposed methods demonstrated superior correlation with actual methylation heterogeneity compared to existing methods.
• Methylation heterogeneity provides distinct biological information beyond conventional methylation levels.
• Distinct methylation heterogeneity profiles in Arabidopsis predicted regulatory roles between genomic elements.
• The methylation heterogeneity score shows potential for identifying loci in human cancers for early detection.

Conclusions:

• Novel model-based methods (MeH) were developed to analyze genome-wide DNA methylation heterogeneity.
• These methods offer a valuable tool for monitoring cellular heterogeneity.
• The findings open new avenues in plant epigenomics and cancer biomarker discovery.