Fig. 9

Motif structure controls variability in expression a Individual nucleus fluorescence levels driven by the construct dst6x. Expression data on each embryo was scaled to minimize the the sum of squared pairwise differences in fluorescence intensity between embryos in 1% bins along the AP axis. Points are colored according to their local density. b Expression levels driven by the construct zld6x, plotted as in panel A. c To test whether differences in standard deviation can be explained by different mean expression levels, the standard deviation as a function of the mean expression is plotted for each 1% bin along the AP axis for both dst6x (red) and zld6x (blue). The relationship between standard deviation and mean is linear with different slopes for each construct. d Two distributions of mRNA count, divided by the mean, in a stochastic transcription model in which the number of transcripts and the ON-OFF state of the promoter are coupled random variables (see Methods). The parameter N represents the strength of transcription when the gene is in the ON state, and b is the probability of finding the gene in the ON state. e Fluorescence values of nuclei were divided by the mean levels in each 1% bin from 60 to 80% embryo length. The resulting distribution in fluorescence about the mean for dst6x is shown. The distribution with N=20 and p=0.5 from panel D is also shown (black line). e The distribution of fluorescence about the mean for zld6x is shown. There is a significant difference in the variance of the zld6x and dst6x distributions (Fligner-Killeen test, p=1.7×10⁻⁸). Additionally, the distribution with N=11 and p=0.9 from D is shown (black line)