Fig. 3

HighMT cells present varied distribution in spatial transcriptomics analyses of breast carcinoma and lung adenocarcinoma. a H&E staining of breast ductal carcinoma in situ (DCIS) analyzed with Visum HD. b H&E image overlay showing the annotated cell types. We use the log1p-normalized gene expression in cells segmented using bin2cell to perform Leiden clustering to define clusters, each aggregated into a “metacell” (Methods). Four primary cell type categories are identified, with copy number variation distinguishing malignant from healthy cells. c UMAP representation of the “metacells” in the tissue with cell type annotations. d Distribution of the pctMT (% MT counts) across cell types in bin2cell-estimated cells, analyzed by a Mann–Whitney U test (****: p < 0.0001). The plot is clipped at the 25% mark on the y-axis to better visualize the differences in distributions. e H&E image overlay showing median mitochondrial count percentages of malignant spots in the 1000 × 1000px patches. Regions with too few malignant cells are excluded. Breast regions of interest are marked as Br.A, Br.B, and Br.C. f Cell type annotations, pctMT values, and H&E staining of cells in regions of interest, with cell type annotations derived from metacell data. g–l Same analyses as a–f for lung adenocarcinoma (LUAD)