Extended Data Fig. 2: Isolation and scRNA-seq quality control of human CD34+ HSPCs.

(a) Representative sorting scheme for isolation of CD34⁺ HSPCs from fetal liver (FL), umbilical cord blood (CB) and postnatal bone marrow (BM) specimens. (b) Summary of scRNA-seq quality control data for each donor/library included in this study indicating unique genes per single cell (top), unique molecular identifiers (UMIs) per cell (middle), and percentage mitochondrial reads (bottom). (c) Single cells from each of the indicated donors are highlighted in red and projected within UMAP space for the entire dataset. (d) Expression levels of fetal-associated genes IGF2BP3, HMGA2, and LIN28B are displayed as violin plots for all cells within the seven age ranges of human life profiled in this study. The horizontal line in each violin plot indicates mean expression level. Significance of all pairwise comparisons between age ranges is also displayed. All p-values were calculated using the two-sided Wilcoxon rank-sum test followed by Bonferroni correction for multiple (28 total) comparisons. All adjusted p-values >0.05 are in shades of blue, =0.05 in white, and <0.05 in shades of red. (e) The percent of X and Y chromosome genes present in each of the indicated gene sets and top 50 contributing genes to each of the indicated gene sets is displayed. Marker genes for each age range were identified through differential expression of all cells from that age range versus the rest of the full dataset. (f) kBET analysis yields no statistically significant difference (p-value >0.05 using the kBet implementation of Pearson’s χ² test across independent library preparations of 2 groups of donors) between Ideal/Expected rejection rate for our dataset and the rejection rate of Seurat batch correction of our human lifetime dataset. Rejection rate and 95% confidence interval for rejection rate are indicated. Additionally, note that in our dataset there is a peak in lymphoid progenitor cell contribution to total hematopoiesis during childhood.