Fig. 1
From: Fast and robust adjustment of cell mixtures in epigenome-wide association studies with SmartSVA
An overview of the simulation strategy. The distributions of cell proportions and methylation values are simulated to reflect the real data with various sources of variability. a The flow chart of the simulation process. b The distribution of the leukocyte subtype proportions generated using a Dirichlet distribution. c The bimodal distribution of methylation beta-values generated using a three-component mixture model. d Simulated batch effects cluster the samples into distinct groups. Principal component analysis (PCA) on the methylation beta-values was used to project the samples onto the first two principal components (PCs). e Correlation between two biological replicates reflecting the variability due to various sources