MethylSig

MethylSig analyzes genome-wide DNA methylation differences from whole-genome bisulfite sequencing (bis-seq), reduced representation bisulfite sequencing (RRBS), and enhanced RRBS to identify significant CpG methylation changes between biological groups.

Key Features:

• Statistical Analysis Framework: MethylSig employs a beta-binomial model to account for read coverage and biological variation across CpG sites or regions.
• Incorporation of Local Information: It enhances estimation of group-level methylation levels and variances by incorporating local information, which is beneficial for small sample sizes.
• Calibrated Type-I Error Control: Permutation studies based on enhanced RRBS samples demonstrate well-calibrated type-I error rates for datasets with three or more samples per group.
• High Sensitivity: Simulations indicate higher sensitivity compared to several alternative methods for detecting methylation differences.
• Single-CpG Resolution: The method supports analysis at single CpG site resolution.

Scientific Applications:

• Gene regulation and cellular specification: Useful for research into DNA methylation roles in gene regulation and cellular specification.
• Precise methylation profiling: Enables single-CpG resolution analyses required for studies demanding precise methylation profiling.
• Comparative epigenomic analyses in disease: Applied to comparative analyses such as distinguishing subtypes of acute leukemia from normal bone marrow samples.

Methodology:

MethylSig fits a beta-binomial model that integrates read coverage with biological variability, incorporates local information for group-level methylation and variance estimation, and uses permutation studies on enhanced RRBS samples to assess type-I error calibration.