Appendix 3—figure 2. Generating a null distribution of correlation coefficients for the anatomical vs. transcriptional similarity.

Correlations are significant relative to null distribution (shown in green) generated by randomly sampling subsets of 2835 homologous genes corresponding to the biologically informed subsets, recomputing the transcriptional similarity, and correlating that with the anatomical similarity 10,000 times. We have subsets of 91 genes in (A) corresponding to X-linked genes, 34 genes in (B) corresponding to sex hormone genes, 11 genes in (C) corresponding to androgen games, and 23 genes in (D) corresponding to estrogen and progesterone genes. For each, the observed correlation was compared to the null correlation to generate a p-value, displayed on the graph.