HMM-DM: identifying differentially methylated regions using a hidden Markov model.

Area of Science:

• Epigenetics
• Genomics
• Bioinformatics

Background:

• DNA methylation is a key epigenetic modification regulating gene expression, crucial for development and differentiation.
• Identifying differential DNA methylation patterns aids in understanding genomic regions linked to various diseases.
• Bisulfite sequencing (BS) technology enables single-CpG resolution methylation analysis, but efficient statistical methods for region detection are lacking.

Purpose of the Study:

• To develop and validate an efficient statistical method for identifying differentially methylated (DM) regions in bisulfite sequencing data.
• To address the limitations of existing methods in detecting DM regions by accounting for spatial correlation and sample variation.

Main Methods:

• A novel approach, HMM-DM, was developed using a hidden Markov model (HMM).
• HMM-DM first identifies differentially methylated CpG sites, considering spatial correlation and inter-sample variability.
• Identified DM sites are then aggregated into differentially methylated regions.

Main Results:

• HMM-DM demonstrated superior performance compared to the BSmooth method in simulation studies.
• The method effectively identifies differentially methylated regions from bisulfite sequencing data.

Conclusions:

• HMM-DM provides an efficient and accurate approach for detecting differentially methylated regions in BS data.
• This method has practical applications in studying disease-associated genomic alterations, as shown in a breast cancer dataset analysis.