Fig. 3. Neighborhood enrichment analysis of human cell types.

a Neighborhood enrichment analysis for neuronal subclasses of all three tissue sections (shown as mean, n = 3 tissue sections). Cluster proximity of subclasses is quantified with a permutation-based test (1000 permutations), by comparing the actual cell type labels with a random configuration of labels and maintaining the positional information. From each pair (actual label vs. label of each permutation), the means and standard deviations are estimated, and a Z-score is calculated. The Z-score indicates if a cluster pair is over-represented or over-depleted in the analysis. b Maps on top of density plots (black) for subclasses from neighborhood enrichment analysis in a, highlighting the increased spatial proximity of the subclasses.