Fig. 3: Correlation of different characteristics of the gene expression profiles.

Max–min is the difference between maximal and minimal expression values. Dots represent one subject evaluated over one day (analysis assumes independence between days; d–f, j–l, p–r), or nine mouse peripheral tissues (each a 46-h time series; a-c, g-i, m-o). a–f Circles depict the relation between max-min and mean for Per2 (a) and PER2 (d), Arntl (b) and ARNTL1 (e), and Nr1d1 (c) and NR1D1 (f). Linear regression suggests a correlation between max-min and mean (p < 0.001). Slopes match between human (violet) and mouse (blue, dashed, and straight lines are the same) in d, f, while for e, treating the three largest values of ARNTL mean levels as outliers result in a correlation line (dashed violet line) closer to the mouse. The regression slope is lowered when restricting the mouse data to the human interval (green circles and lines), see methods. g–l Correlation of gene expression amplitude and maximum expression. Circles depict the relation between max-min and max for Per2 (g) and PER2 (j), Arntl (h) and ARNTL1 (k), and Nr1d1 (i) and NR1D1 (l). Linear regression suggests a correlation between max-min and max (p < 0.001). m–r: Circles depict the relation between the mean expression of different clock genes. Linear regression suggests a correlation between all mouse genes (m–o, p < 0.002), for humans, mean NR1D1 expression correlates with mean PER2 expression (q).