MethylHMM

MethylHMM analyzes in vivo double-stranded DNA methylation patterns to infer DNA methyltransferase properties such as processivity and substrate specificity.

Key Features:

• Hidden Markov Model (HMM): Fits a novel Hidden Markov Model to observed methylation data to model methylation state transitions.
• Bisulfite conversion error modeling: Explicitly accounts for potential bisulfite conversion errors in the methylation observations.
• Statistical estimation of enzymatic parameters: Uses statistical methods to produce quantitative estimates of enzyme processivity and substrate specificity.
• Hairpin-bisulfite PCR data support: Analyzes in vivo double-stranded CpG methylation patterns obtained using hairpin-bisulfite PCR.
• Multi-species and locus application: Applied to human loci FMR1, G6PD (Xi-linked), and LEP (autosomal) and to mouse DNMT1 activity in vitro for comparative analysis.

Scientific Applications:

• DNMT1 processivity assessment: Provides evidence for long DNMT1 association tracts at human FMR1 and G6PD loci, reaching several hundred base pairs.
• Substrate specificity characterization: Quantifies DNMT1 preference for hemi-methylated CpG sites over unmethylated CpGs across studied loci.
• DNMT3 involvement inference: Indicates minimal contribution of de novo DNMT3 enzymes at the examined loci under the studied conditions.
• Comparative epigenetic analysis: Enables comparison of DNMT1 activity between human and mouse data, revealing shorter association tracts in mouse data.

Methodology:

Fits a Hidden Markov Model to observed methylation data while modeling bisulfite conversion errors and uses statistical estimation to infer enzyme processivity and substrate specificity.