Fig. 4: stClinic evaluates the malignancy level of niches on the 43 TNBC slices.

a Kaplan–Meier survival curve for 43 breast cancer slices, with slices stratified into low-risk and high-risk groups based on their median overall survival time. Unadjusted two-sided log-rank test. b UMAP visualization of the latent features by stClinic on 43 slices, where the colors in the left and right panels indicate slices and clusters, respectively. c Heatmap showing the cell-type proportions on spatial domains by stClinic. d H&E plot of slices 15 and 26. Slice 15 was annotated with fibrosis, necrosis, tumor, and dense lymphoid infiltration regions, while slice 26 was annotated with fibrosis and tumor regions. e Spatial domains identified by stClinic on slices 15 and 26. f Kaplan–Meier survival curves for 43 breast cancer slices, with slices classified into low-risk and high-risk groups based on their median hazard ratio predicted by stClinic. Unadjusted two-sided log-rank test. g Bar plot displaying the weights of cluster in prognosis by stClinic. h Heatmap depicting the average gene expression of the top 30 over-expressed genes for clusters 10 and 6. i Overall survival rate of patients with the low or high expression patterns of the top 30 over-expressed genes for clusters 10 (top panel) and 6 (bottom panel) in breast cancer data from TCGA by GEPIA2⁸⁹. Unadjusted two-sided log-rank test. j Functional enrichment analysis of the over-expressed genes in cluster 10. Unadjusted one-sided Fisher’s exact test. k Boxplot showing the total copy number gains and losses per spot in clusters 10 (N = 2717) and 6 (N = 5124), as inferred by CopyKAT⁹⁰. Unadjusted one-sided Wilcox rank-sum test. 2.22e-16 represents a very small number, effectively close to zero. For each boxplot, the center line, box limits, and whiskers separately indicate the median, upper and lower quartiles, and 1.5 × interquartile range. Source data are provided as a Source Data file.