genes than other clusters. For the 16 synaptic gene sets identified in cluster 1, the fraction of genes that were downregulated in each set was significantly higher than that for the remaining 16 gene sets (mean fraction of downregulated genes was 0.712 and 0.507, respectively; t = 6.26; df = 30; p = 6.7 × 10−7). Furthermore, we found that there were significant differences in the mean fraction of downregulated genes across the four clusters (analysis of variance, F[3,28] = 28.46, p = 1.19 × 10−8). The mean fraction of downregulated genes for clusters 1, 2, 3, and 4 was 0.76, 0.63, 0.69, and 0.38, respectively (Fig. 2; Table S3): across all genes contained in cluster 1, there was a significantly higher fraction of downregulated than upregulated genes (t = 4.91, Bonferroni corrected p = 1.34 × 10−4), while cluster 4, containing pathways involved in receptor tyrosine kinase activity, circulatory system development, and plasma membrane regulation, had a significantly higher fraction of upregulated genes (t = −7.4, Bonferroni corrected p = 4.94 × 10−7). Clusters 2 and 3, representing cytosolic plasma membrane components and transmembrane transport, respectively, had more similar levels of up and downregulated genes.
