eDMR

eDMR identifies differentially methylated regions (DMRs) from enriched whole-genome bisulfate conversion sequencing data using a bimodal normal distribution model and a weighted cost function, implemented in R.

Key Features:

• Bimodal Distribution Model: Uses a bimodal normal distribution model to determine boundaries of regions with significant epigenetic modification.
• Weighted Cost Function: Implements a weighted cost function that incorporates the spatial distribution of CpG sites for regional methylation analysis.
• Empirical DMR Identification: Extends the methylKit R pipeline to empirically determine DMRs from high-throughput enriched whole-genome bisulfate conversion sequencing data.
• Dependent p-value Adjustment: Performs dependent adjustment for combination of regional p-values to enhance statistical robustness of DMR calls.
• DMR Annotation and Classification: Annotates identified DMRs and classifies their directionality and genome-wide distribution, enabling biological interpretation and subtype stratification such as Acute Myeloid Leukemia (AML) tumor sub-types.
• High-throughput Compatibility: Optimized for analysis of large-scale methylation sequencing datasets from enriched whole-genome bisulfate conversion sequencing.

Scientific Applications:

• Epigenomics Research: Facilitates identification and characterization of genomic regions with altered DNA methylation during development or disease.
• Clinical Stratification: Supports classification of DMRs for clinical applications, including stratifying tumor sub-types such as AML.
• High-throughput Methylation Sequencing: Applied to large-scale methylation sequencing experiments to detect significant epigenetic modifications from enriched whole-genome data.

Methodology:

Applies a bimodal normal distribution model, a weighted cost function incorporating CpG spatial distribution, dependent adjustment for regional p-value combination, and extensions of the methylKit R pipeline for empirical DMR identification and annotation.