Fig. 3: Data-driven anatomical regions and borders.
From: Scalable in situ single-cell profiling by electrophoretic capture of mRNA using EEL FISH
a, Clustering of hexagonally binned expression data results in distinct anatomical regions. b, Local transcriptional heterogeneity can be quantified to obtain border strength and angle. c, Transcript densities for selected genes show layer-restricted expression between the pia and white matter, where interfaces between domains are marked by local maxima in border strength. P, pia; WM, white matter. d, The striatum has no clear subregions but contains spatial gradients of certain genes. Mixing region label colors by probability of cluster identity displays these gradients better than discrete colors. Bar plots show mean expression ± s.d. for 2,163 bins. e, Mixed cluster colors for the full sample show gradients as well as discrete borders. f, Mixed region colors rendered with border strength as height to show the difference between sharp borders and gradients.
