Identification of leukemia-enriched signature through the development of a comprehensive pediatric single-cell atlas

Single-cell transcriptome profiling enables unparalleled characterization of the heterogeneous microenvironment of pediatric leukemias. To facilitate comparative analyses and generate pediatric leukemia signatures, we collect, process, and annotate single-cell data comprising over 540,000 cells from 159 different pediatric acute leukemia (myeloid, lymphoid, mixed phenotype lineages) and healthy bone marrow (BM) samples, profiled in our lab and curated from publicly available studies. The analysis identifies a leukemia-enriched signature of nine genes with over-expression in leukemic blast compared to healthy BM cells. This signature is also consistently over-expressed in leukemia samples compared to normal BM in bulk RNA-seq datasets (over 2000 samples). Outcome-based analysis on diagnosis samples using measurable residual disease (MRD) status depicts a significant association of oncogene-induced senescence and g-protein activation pathways with MRD positivity. MRD positivity across pediatric leukemias is also correlated with significant depletion of CD8+ and CD4+ naïve T-cells and M1-macrophages at diagnosis. To enable easy access to this comprehensive pediatric leukemia single-cell atlas, we develop the Pediatric Single-cell Cancer Atlas (PedSCAtlas, https://bhasinlab.bmi.emory.edu/PediatricSCAtlas/ ). The atlas allows for quick exploration of single-cell data based on genes, cell type composition, and clinical outcomes to understand the cellular landscape of pediatric leukemias.