Extended Data Fig. 1: Features of leukemia stem and progenitor cell populations from scRNA-seq.

A) UMAP and PCA embeddings of 4163 AML LSPCs after feature weight derivation with the Self-Assembling Manifolds (SAM) algorithm. B-F) Diffusion map of re-annotated LSPC populations using SAM-derived feature weights, depicting: B) patient identity, C) prior cell type annotation, D) enrichment of LSC-specific genes from Ng et al (2016) and Shannon Diversity Index, E) scaled CDK6 expression and enrichment of the E2F3 regulon, and F) enrichment of E2F1 and CTCF regulons. G) Cell cycle status of Quiescent (n = 1855), Primed (n = 1240), and Cycling LSPCs (n = 1068). H) Enrichment of inflammatory signaling pathways and regulons in LSPCs. I) Transcription factor regulon activity, inferred through pySCENIC, specific to each LSPC. J) Normalized confusion matrix depicting classifier accuracy of prior and new cell type annotations for primitive AML cells. The classifier was built using SingleCellNet, an Ensemble classifier for scRNA-seq data trained from the top pairs of genes unique to each cell type. 800 cells from each cell type were used for training and 250 were used for validation.