Fig. 1

Hierarchical clustering and principal component analysis (PCA) of the Illumina 27K data. The figure is based only on autosomal loci. a Supervised analysis testing for differences between groups using ANOVA and applying a cutoff of <5 % false discovery rate (FDR), corresponding to a p value of 3.11E-4 and 148 CpG sites; b Unsupervised hierarchical clustering on the normalized data without applying a statistical hypothesis. The heat maps in the upper part of the figure correspond to one sample for each column and one locus for each horizontal line; red to green corresponds to relatively hypermethylated (red) to hypomethylated (green). In the lower part of the figure, each of the spheres corresponds to one sample and the connecting lines correspond to the two nearest neighbors (more similar) of a particular sample