Fig. 7: Basal cell subtype discovery in MTECs single-cell transcriptomics data (N=6009 cells).
From: Heterogeneity-preserving discriminative feature selection for disease-specific subtype discovery
UMAP visualizations using PHet’s selected features (a), pre-annotated markers (b), and HV features (c). Coloring represents the detected basal cell types. PHet’s selected features and pre-annotated markers are provided as a Source Data file. Dot plots of the expression levels of selected basal and secretory cell markers among the four basal subtypes for PHet (d) and pre-annotated markers (e). Violin plots of the distribution of z-scaled feature expressions for selected basal and secretory cell markers, indicated on the y-axis labels, among the four basal cell subpopulations represented on the x-axis labels for PHet (f) and pre-annotated markers (g). Silhouette scores (h) for clustering results based on PHet’s features, pre-annotated markers, and HV features. The cluster quality results are provided as a Source Data file. UMAP plots of basal cells for PHet (i) and pre-annotated markers (j). These plots provide insights into the expression patterns of the secretory cell marker BRIFA1. UMAP visualizations of injured vs uninjured conditions using PHet’s selected features (k), pre-annotated markers (l), and HV features (m). The proportion of each basal subtype in the injured and uninjured cell groups given PHet (n), pre-annotated markers (o), and HV features (p).
