Extended Data Fig. 6: Marker genes and cell type–specific TF regulator activity for sub-clustered interfollicular and hair-follicle associated keratinocytes.
(a) UMAP projections of sub-clustered keratinocyte scRNA-seq dataset colored by expression levels of representative marker genes. (b) UMAP projections of sub-clustered keratinocyte scATAC-seq dataset colored by gene activity scores of the same marker genes shown in (A). (c) Heatmap showing the chromatin accessibility (left) and gene expression (right) for 28,991 keratinocyte-specific peak-to-gene linkages. Peak-to-gene linkages were clustered using k-means clustering (k = 12). Rows indicate peak accessibility and gene expression on the left and right heatmaps respectively. Each column is a pseudo-bulk sample, with the colorbar on top of each heatmap indicating the cluster identity of each pseudo-bulk sample. (d) Hypergeometric enrichment p-values of TF motifs in peaks from each of the k-means clusters from (C). (e) Plot of TF motif activity correlation to corresponding TF gene expression across sub-clustered dataset against the maximum difference in chromVAR deviation z-score between clusters. TF’s with a maximum chromVAR difference in the top quartile and a pearson correlation greater than 0.5 are colored in red. (f) Prioritization of gene targets for LHX2. The x-axis shows the Pearson correlation between the TF motif activity and integrated gene expression for all expressed genes across all keratinocytes. The y-axis shows the TF Linkage Score (for all linked peaks, sum of motif score scaled by linkage correlation). Color of points indicates the hypergeometric enrichment of the TF motif in all linked peaks for each gene. Top gene targets are indicated in the shaded area (motif correlation to gene expression >0.25, linkage score >80th percentile). GO term enrichments for the top gene targets are shown in the inset bar plot. (g) Same as in (F), but for androgen receptor (AR). (h) Same as in (F), but for POU2F3.
