24-h rhythmic genes

24-h rhythmic gene set consist of genes (probe sets) whose expressions of PT are significantly oscillated both in short-day and long-day conditions.

Method to identify 24-h rhythmic genes

For identifying 24-h rhythmic genes, we used natural expression values from the GeneChip data of short-day and long-day conditions. We concatenated two time-courses of experimental replicates (6 time points for each replicate) and used 12 time-point dataset. To identify 24-h rhythmic genes in the mouse PT, we first calculated the best Pearson's correlation between gene expression data and cosine curves of a 24-h period, and determined its phase in short-day and long-day conditions using a Fourier transformation-based method [14]. We then calculated the P-values for the best correlations of the short-day and long-day using random shuffling, and then combined these two P-values with Fisher's method. We estimated the false-discovery rates (FDRs) of the combined P-value based on the Benjamini-Hochberg method, and selected significantly oscillating genes as those for which the FDRs were less than or equal to 0.05.