Fig. 3

Pi-ATAC dissects EpCAM+ tumor cells and tumor-infiltrating immune cells from the same mouse breast tumor. a Schematic illustrating the different cell types in the mouse breast tumor. b t-SNE projection of TF deviation z-scores of Pi-ATAC EpCAM+ (n = 177) and CD45+ (n = 192) cells from the same mouse breast tumor. c Unsupervised hierarchical clustering of the TF deviation z-scores of all 84 significant variable TFs (p < 0.05 after Benjamini–Hochberg (BH) correction on multiple tests) across EpCAM+ (n = 177) and CD45+ (n = 192) cells from the same mouse breast tumor. Each column represents one cell and each row a transcription factor motif. Motif modules (m1–3) and cell subgroups (s1–7) are marked with distinguished colors. In addition, the staining cluster information from FACS was assigned to each individual cell (top color bar). d t-SNE projection of TF deviations of 84 significant variable TF motifs of EpCAM+ and CD45+ cells isolated from a tumor, color coded by cellular subgroup information; e-g color coded by the accessibility of the TF motif with most significant variability in each module (Fig. 3d): Ets2 in m3 (e), Smarcc1 in m2 (f) and Hif in m1 (g). Motifs are based on PWM from CisBP database. Red is highly accessible, blue is low accessible; h color coded by immune-phenotype. i Scatter plot of TF motif variability calculated by ChromVAR across 4T1–splenocyte mixture to TF variability calculated by ChromVAR across EpCAM+ – CD45+ primary tumor cells. Colors indicate TF motif modules as in (c). Arrow points to Hif