Fig. 3: Lineage dissection of the human mammary scRNAseq data and the cells of origin of breast cancer.

a The UMAP projection of the human normal breast epithelial (Individual #4) clustered and annotated based on the definitive marker gene expressions or the label transfer using the integrated mouse data as a reference. b The agreement between the definitive annotation and the label transfer for the human normal breast epithelium from the four individuals. The progenitor and differentiated clusters from the same lineage in the mouse were merged and the cells are partitioned by the definitive and transferred annotation. The point size indicates the percentage and the sum of each column equals to 100. c The workflow of the lineage trajectory inference and the data integration (4 individuals, N = 24,377) based on the human lineage-specific gene sets with the scGSVA and the UMAP dimensionality. The “Stem” gene set was adopted from the mouse data. d, e The UMAP projection and the ternary plot based on the four (“Stem,” “Basal,” “Alv,” and “Hor”) and three (“Basal,” “Alv,” and “Hor”) gene sets, respectively, for the TCGA breast cancer RNAseq data (d) and the human breast cancer scRNAseq data (e).