Fig. 2: Cellular neighborhood structures and unique spatial niches in B cell lymphoma.

a, High-resolution cell state identification in the CosMx SMI dataset. b, Schematic depiction of cellular neighborhood analysis, created with Biorender.com. A cellular neighborhood of 200 pixels in radius was identified for each cell. On top of that, the neighborhood compositions, spatial niches and neighborhood-based cell–cell communication analysis were performed. c, The neighborhood compositions of 19 different cell states in the CosMx SMI dataset, as represented by the sub-segmented barplots colored according to panel a. d, Identification of seven unique spatial niches from unsupervised clustering of cells based on cellular neighborhood compositions. The distribution of cells from spatial niches in the PCA space is shown. e, The cell state compositions of each spatial niche, represented by the sub-segmented barplots colored according to panel a. f, The representative DEGs in each spatial niche. g, Representative FOV images from CosMx SMI, each with a dominant spatial niche. CN cluster (above) is colored according to panel e, and cell type (below) is colored according to panel a.