This representation (see Fig. 1c) was generated using 1 − r as a distance metric, where r is the pairwise Pearson’s correlation coefficient calculated across all gene expression probes for each pair of samples. These distances were coupled with an MDS algorithm to attempt to satisfy all the pairwise distances in two-dimensional space. For both MDS and PCA, three-dimensional analyses more precisely depict systematic age effects (Supplementary Fig. 1, parts 2 and 3). It is important to note in this analysis (and those shown in Figs 1b and 4) that because expression data are expressed as a ratio to reference here, the mean expression correlation across all samples is near zero: r = 0.02.
