A cell-state coordinate system constructed from scRNA-seq measurements of HSPCs. (A) PCA of native HSPC scRNA-seq data with cells colored by genotype reveals differences in density across selected gene expression space. (B) PCA plots with cells colored by expression levels of different gene markers (Mecom, HSC; Pf4, megakaryocytes; Mpo, granulocytes; Dntt, lymphocytes). (C) Data (WT HSPCs) used for pseudotime analysis colored by input clusters defined by listed marker genes, with resulting differentiation trajectories given by black arrows. ErP, erythrocyte-primed; MkP, megakaryocyte-primed; MyP, myeloid-primed; LyP, lymphoid-primed. (D) Projection of published sorted HSPC scRNA-seq (inDrops platform) data⁵⁰ into same PCA space (left). Jensen Shannon Divergence (JSD) values (right) between WT HSPC cell-identity distributions (lineage weights) and local distribution of cell types in sorted HSPCs within radius specified (dark gray). Most JSD values are closer to zero demonstrating similarity between the cell type and lineage-weight distributions. As a reference, the same computation was performed on the sorted HSPCs, using the sorted cell types as their cell-identity distributions (light gray). (E) PCA of native HSPC scRNA-seq data with cells colored by predicted cell type from classification model³⁰ built from microarray data of sorted HSPCs.²⁵ (F) Fraction of each cell type defined by maximum lineage-weight predicted as each cell type by classification model demonstrates strong correspondence (59%-68% match). (G) Logarithm of the ratio of IκB⁻ to WT HSPC density over pseudotime for the different lineages in both experimental replicates. Densities were estimated 1000 times to reflect uncertainty in lineage assignments; solid line indicates median and shaded region covers minimum to maximum values.