RaceID

RaceID is a computational method for identifying rare cell types in single-cell transcriptomic datasets, designed to overcome limitations of approaches that primarily resolve abundant populations or rely on a small set of known marker genes. It analyzes transcriptomes from complex cell mixtures and detects discrete subpopulations that may be represented by extremely few cells—down to a single cell in a randomly sampled population—enabling unbiased discovery of previously unrecognized cell states.

In its initial application to mouse intestinal organoids, RaceID resolved cell types spanning a wide abundance range and enabled the identification of Reg4 as a marker for rare enteroendocrine cells. Using Reg4-based enrichment, the approach supported deeper characterization of enteroendocrine heterogeneity, recovering known lineages and revealing additional subtypes that were subsequently validated in vivo. Applied to ex vivo intestinal stem-cell preparations, RaceID further distinguished a dominant, homogeneous population of Lgr5-positive stem cells from a rare subset of Lgr5-positive secretory cells. Overall, RaceID provides a general framework for rare-population discovery and marker-gene identification in healthy and diseased tissues.