Fig. 1: A spatially resolved atlas of human breast cancer slides across different molecular subtypes and tumor sites.

a Schematic diagram illustrating the acquisition and analysis workflow of the ST maps. Stereo-seq was performed on 30 slides of primary breast tumors and paired lymph node metastases from 23 patients, yielding a total of 4,894,305,510 DNBs. ScRNA-seq was performed on the fresh tumor samples from 6 out of 23 patients and identified 78,848 qualified cells. Each block represents a ST slide and is colored by molecular subtype. Red, blue, orange, and green spots represent venular endothelial cells (vECs), venular smooth muscle cells (vSMCs), tumor cells, and immune cells, respectively, in the projection of scRNA-seq and ST data. Schematic plots were created using bioRender. b Uniform manifold approximation and projection (UMAP) plot of the scRNA-seq dataset. Each cluster is colored by its corresponding cell type. EC: endothelial cell; SMC: smooth muscle cell, including vascular SMC and pericyte; T/NK: T or natural killer cell. c Differential expression of canonical markers in each cell type cluster. Macro: macrophage; Fibro: fibroblast; Melano: melanocyte; Granulo: granulocyte; PB: plasmablast. d–f Hematoxylin and eosin (H&E) staining (d), gene count maps (e), and cell type annotation maps (f) of the ST slides of the primary tumor (PT) and metastatic lymph node (LN) from the patient NCC-P05. ST spots are mapped to the H&E staining of adjacent slides. Raw spatial expression matrix of each bin/spot is convoluted into pseudo 50 spot with 25 µm squares (50 × 50 bins/spot, bin50 spot for short). IDC: invasive ductal carcinoma; DCIS: ductal carcinoma in situ. g The cell type annotation, H&E staining, gene counts, and unique molecular identifiers (UMI) counts in the selected area of NCC-BR5 slide highlighted in Fig. 1d–f. h Relative proportions of cell types across 30 ST slides. ST slides are grouped according to the molecular subtype and tumor site.