Fig. 3. stMVC enables the detection of cell-states distributed in different spatial domains on ovarian and breast cancers.

a Immunofluorescent staining of the tissue section and 17 manually segmented tumor regions. The intensity of DAPI, cytokeratin, and CD45 is shown in blue, green, and yellow. b Spatial clustering by DR-SC, Squidpy, stLearn, BayesSpace, STAGATE, and stMVC. c UMAP visualization of latent features by stMVC. Each domain is outlined by the region it belongs to, i.e., tumor or non-tumor. d Heatmap of average gene expression of signature genes from 16 domains by stMVC. Rows and columns indicate genes and domains. e UMAP plot of average expression of signature genes for domains 11 and 12 in scRNA-seq data of 4081 epithelial cells from one independent ovarian cancer. f Enrichment of infiltrating stromal and immune cells in each domain compared to the total distribution of those in four domains. The ratio is calculated by chi-square test⁸⁵. The larger the ratio, the more cells are enriched in the domain. g Total survival rate of patients with the high ($>$ median value) or low (low $<$ median value) expression level of 11 signature genes for domains 10 and 11 in gene expression data of ovarian cancer from TCGA by GEPIA2⁸⁶ (Supplementary Table 1). h Immunofluorescent staining of the tissue section and 15 manually segmented tumor regions, where those regions are outlined by their histological annotations: invasive carcinoma (red), carcinoma in situ (yellow), and benign hyperplasia (green). The intensity of DAPI, fiducial frame, and anti-CD3 is shown in green, blue, and yellow. i Spatial clustering by DR-SC, Squidpy, stLearn, BayesSpace, STAGATE, and stMVC. j UMAP visualization of latent features by stMVC. Domains are outlined by their histological annotations. k Heatmap of gene expression of signature genes from five domains enriched in the carcinoma in situ region by stMVC. Rows and columns indicate genes and domains. l Spatial expression of classical marker genes for stromal and immune cells, including PTPRC, CD14, IGCL2, IGHG3, MS4A1, MYLK, HLA-DRB1, CD4, CD8A, ENG, FABP4, and FCGR3B for data denoised by stMVC. Source data are provided as a Source Data file.