Figure S1.

CITE-seq analysis of the LSK cells at steady-state. (A) Differentially expressed genes in each LSK subpopulation based on CITE-seq analysis with ICGS2 (column = cells, rows = genes). The top marker gene (MarkerFinder) for each assigned cluster (right) is shown for each LSK cluster (left). (B) Clustered heatmap of intra-correlated (rho > 0.3) cell cycle genes for cells corresponding to panel A. Cell cycle phase determined by gene set enrichment against PathwayCommons. (C) UMAP visualization of marker genes corresponding to HSC, megakaryocytic, myeloid, and lymphoid markers/regulators. (D) Upper panel: Pseudotime-based approach, Slingshot, on the UMAP plot (Fig. 1 A) shows the predicted myeloid (red line), lymphoid (yellow and pink lines), and erythroid-megakaryocytic (blue and green lines) lineage trajectories. C1 HSC cluster was set as the starting point (indicated by green dot) toward the predicted differentiation states. Lower panel: Cell-to-cell random walks calculated from RNA velocity using CellRank are shown on the same UMAP plot shown in the upper panel. C1 HSC cluster was set as the starting point for lineage inference or the simulation of cell-to-cell random walk (as indicated by the black dots). The terminal points of random walks, or lineage development, from the starting point are indicated by the yellow dots. Random walks are colored according to the pseudotime from the starting to terminal points of differentiation. We notice that the segments in C1 HSC clusters indicate early stage of differentiation and the segments in clusters C9, C10 (myeloid), C6, C11 (lymphoid), and C3 (erythroid-megakaryocytic) indicate later stages of differentiation. (E) Heatmap of the top 60 scRNA-seq LSK marker genes for ICGS2 clusters viewed through bulk RNA-seq from Cabezas-Wallscheid et al. (2014) (independently median normalized for each). (F) Estimated cell type fraction in bulk RNA-seq from Cabezas-Wallscheid et al. (2014) as determined by CIBERSORTx. (G) TF-to-gene pairwise correlation heatmap (HOPACH clustering) based on scRNA-seq gene expression in three MPP4 predicted cell populations (C6, C7, and C11). Red = positive Pearson correlation, blue = negative Pearson correlation. In E and F, HSC and MPP1 defined as LSK Flt3⁻ CD48⁻ CD150⁺ CD34⁻ and LSK Flt3⁻ CD48⁻ CD150⁺ CD34⁺, respectively.