Statistically significant edges were defined as follows. Each interaction (characterized by the secreting cell type, ligand, receptor, and the receiving cell type) was assigned a score calculated as the sum of the scaled ligand and receptor expression (the edge weight). A P value for each interaction was calculated using a one-sided Wilcoxon rank sum test separately for each species. There were no assumptions made for the statistical distribution of the ligand-receptor product score, and the P value only considered the score in relation to the scores of other interactions in the same species. Last, a P value for the significance of each interaction type across all species was calculated by combining the respective species-specific P values using Fisher’s method. Doing so allows both the fusion of statistical significance evidence from multiple species to boost the overall statistical significance for that interaction and the appropriate scale of the P value if the particular ligand or receptor is missing from one of the species.