GMQN

GMQN corrects technical variation and probe-design biases in DNA methylation data from Illumina HumanMethylation BeadChip 450K and 850K arrays to enable accurate epigenome-wide association studies (EWAS) and downstream methylation analyses.

Key Features:

• Reference-Based Methodology: Employs a reference-based approach to normalize signal intensities across samples, removing unwanted technical variation.
• Batch Effect Correction: Aligns the distribution of methylation signals to correct batch effects arising from different experimental runs or batches.
• Probe Bias Mitigation: Addresses probe design bias to reduce systematic deviations among probes and improve comparability across CpG sites.
• Integration with Other Normalization Methods: Can be combined with Subset-quantile Within Array Normalization (SWAN) and Beta-Mixture Quantile (BMIQ) normalization methods.

Scientific Applications:

• Epigenome-wide association studies (EWAS): Provides normalized methylation data suitable for detecting associations between methylation and phenotypes or exposures.
• Large-scale epigenetic studies: Enhances comparability and reliability when integrating data across many samples or cohorts.
• Differential methylation analysis: Improves accuracy of identifying differentially methylated CpG sites by reducing technical artifacts.
• Integrative genomics and biomarker identification: Facilitates downstream integrative analyses and the discovery of methylation-based biomarkers.

Methodology:

GMQN models methylation signal distributions using Gaussian mixture models and applies quantile normalization to align methylation level distributions across samples.