Fig. 1: Integration of Normal and PDAC Single-Cell RNA-Seq Data.

a Graphical workflow of the single-cell RNA-seq processing pipeline for the discovery set and atlas. A broad collection of both normal and tumor pancreas scRNA-seq data was curated by leveraging publicly available data sets. After removal of outlier cells, the standard Seurat and Harmony processing steps were performed to remove batch effects. b t-Distributed Stochastic Neighbor Embedding (t-SNE) projection of the full dataset showed the relative composition of the atlas based on Patients, Condition, Dataset, and Cell Type [Level 1]. c Canonical cell type marker genes were used to highlight specific expression of EPCAM (Epithelium), AMBP (Normal Duct), MUC1 (Aberrant Duct), COL1A1 (Fibroblasts), PECAM1 (Endothelium), RGS5 (Pancreatic Stellate Cells), AIF1 (Myeloid), MS4A1 (B-Lymphocytes), and CD3D (T-Lymphocytes). d Major cell type compartments (Endothelium, Stroma, Myeloid, Lymphocytes) were processed further to identify cell type [Level 2] subpopulations. e Differential secretome factors (ligands and receptors) expressed by each cell type subpopulation (Log-Fold Change >0.8). Expression groups are organized by the broader Cell Type 1 compartments. Color and size of dots represent % of cells expressing the gene and average expression respectively. Label colors indicate feature categories (TNF-Pathway, Integrins, EGF domain, Interleukin signaling. Source data are provided as a Source Data file.