**Gene specific error model**

The standard error was determined using the data from 43 separate experiments that were each repeated between two and five times. These experiments include 7 experiments reported by Jiang et al., 5 experiments reported by Reinke et al., and 31 experiments that are currently stored in the Stanford Microarray Database. To calculate the standard error (SD) from all of the experiments, the following formula was used:

SD_{total} =
SQRT((n_{x}-1)SD_{x}^{2} +
(n_{y}-1)SD_{y}^{2} +
(n_{z}-1)SD_{z}^{2} +
…)/(N_{total}-1))

where x, y, z represent individual experiments, n is the number of repeats in each experiment, and N is the total number of repeats. The average standard error for each gene can be found in a tab-delimited text file.

To directly compare data between two samples, a Student's t test was used to determine whether the observed level of regulation was significant. To compare data over a series of experiments such as a time course, a one way analysis of variance was used to identify genes that show significant levels of regulation. Each statistical method used the global standard error as an estimate of standard error of the population.