To minimize potential systematic errors, all stages of the experiment were balanced across experimental groups. That is, equal numbers of animals in each group were sacrificed within the same 2-hr time frame each day. For the ACB-shell samples, equal numbers of RNA preparations from the representative groups were processed through the labeling, hybridization, washing and scanning protocols on a given day, ensuring that the sets of samples labeled together were not also hybridized together, using premixes of reagents. However, the CeA samples were done several weeks after the ACB-shell samples, and, for the CeA, the sets of samples labeled together were also hybridized together; which led to a significant batch effect. Therefore, we incorporated ‘batch’ as a co-variate into the statistical analysis of the CeA samples.
