Fig. 3

A graphical summary of p-values in the SI for gene-group intersections. The two matrices display the p-values for the gene-group intersections in graphical form. The matrices correspond to the two tabs in the SI file intersection_p-values.xlsx (although http://go.usa.gov/3kjsH displays the p-values more conveniently). The matrices omit their upper triangle, because they are symmetric. They also omit their diagonal, because it corresponds to the intersections of each gene group with itself. In each matrix, each of the 66 (unlabeled) rows corresponds to a single gene group in the SI. In the SI and in Fig. 3, each gene-group corresponds to an uncorrected cluster p-value p ≤ 0.20. (In contrast, the present, main article confines itself to examining clusters significant at α = 0.05 with p ≤ 4.72 × 10^− 4.) The row order in each matrix follows the SI, where the sort is primarily alphabetical on the TF and then on the p-value. The magnitudes of the Fisher Exact p-values for the gene-group intersections are in shades of gray: black indicates 0.0; white, 1.0; and 50% gray, the threshold for significance at α = 0.05. For the Fisher Exact Test p-values, the threshold is p = 2.77 × 10^− 5 after the present article’s multiple correction; for validating DAVID p-values, p = 0.05. The font for the (unreadable) p-values is 50% gray, making gray on black significant; gray on white, not significant; 50% gray, borderline significant, etc. Each of the 12 white entries in the DAVID matrix indicates a p-value that DAVID censored because the gene-group intersection was too small