Concentrating cell-type-specific transcriptional signatures in bone marrow from interspecies comparisons

Cell-type definition is commonly achieved using marker genes. Because cell types are broadly conserved across evolution, marker genes are identifiable via interspecies comparisons. We generated single-cell RNA sequencing datasets of bone marrow niche and hematopoietic progenitor compartments from four mouse species. Using these data, we developed a strategy that adds conservation of transcriptional levels to existing approaches that identify marker genes using conserved cell-type specificity. The resulting "signature gene" lists contain both well-known and underexplored bone marrow markers. Signature genes capture cell identities and thus can robustly discern homologous cell types in diverse tissues of evolutionarily distant species. Unbiased benchmarking assessments demonstrated that our signature genes are comparable or superior to larger, less-conserved gene lists. Last, we confirm our framework's versatility and robustness using published datasets from another tissue and mammalian order. Thus, combining conserved cell-type specificity and transcriptional levels is a powerful, widely applicable strategy to distill profoundly descriptive signatures.