The upper row (A1–A3) shows data for all cis-elements including counts for multiple occurrences of cis-elements. In the lower row (B1–B3), only unique cis-element counts (see Methods) have been used. Genes were only included if their associated intergenic upstream region was large enough to fully contain the considered upstream interval, i.e., no overlap of preceding genes and the considered promoter segment was allowed. Shown are the mean values and the SEM. Correlation coefficients, r, and associated p-values are given for all raw data pairs (individual genes and associated motif counts) and, in parentheses, for the mean values as they are plotted in the graph. If the shuffled p-value, pS, was not zero, it is reported as well (see Methods). Mean cis-element counts were only plotted if at least 100 genes were detected at the particular breadth of response. Note: With increasing distance of the considered promoter region from the transcription start site, fewer genes were considered as intergenic upstream regions were frequently not large enough (respective gene counts were A1, B1 = 8,597; A2, B2 = 6,537; A3, B3 = 2,737).