MBAmethyl

MBAmethyl analyzes DNA methylation data from high-throughput array-based platforms by modeling probe affinity effects and spatial correlations to improve methylation level estimation.

Key Features:

• Probe Affinity Effect Estimation: Estimates probe affinity effects across multiple samples by pooling information from the same probe to mitigate probe-specific biases.
• Spatial Correlation Utilization: Leverages correlations among neighboring genomic locations to refine methylation level estimates and capture spatially coherent patterns.
• Iterative Smoothing and Correction Techniques: Applies Group Fused Lasso (GFL) for iterative smoothing based on genomic proximity and uses constrained least squares to correct for probe affinity effects.
• Differentially Methylated Region (DMR) Detection: Implements a procedure for detecting DMRs and compares its performance with state-of-the-art approaches on data applications.

Scientific Applications:

• Enhanced Methylation Profiling: Produces more accurate DNA methylation landscapes from array-based data by correcting probe-specific biases and incorporating spatial information.
• Functional Genomic Insights: Enables detection of DMRs to investigate potential epigenetic mechanisms underlying biological processes and diseases.
• Comparative Studies: Facilitates comparative analyses between conditions or populations to identify methylation patterns associated with phenotypes or environmental exposures.

Methodology:

Uses a statistical model that integrates probe affinity effects and spatial correlations, performs iterative smoothing with Group Fused Lasso, corrects probe effects via constrained least squares, and validates the approach through simulation studies.