RFind-sc maps single cells onto biological axes defined by external gene signatures

RFind-sc maps single cells onto biologically interpretable axes using external gene signatures, reproducing reported microglia, stressed hippocampus, and human blood states while improving accuracy.

• Why it matters: Single-cell embeddings like UMAP and t-SNE lack biological meaning, and prior external-reference approaches have limited interpretability for measuring cell states and stress responses.
• What they did: RFind-sc compares each cell to external references and places it on axes defined by expression signatures, gene sets, or genetic perturbations, including ~1,900 in vivo Perturb-seq knock-outs.
• The result: It reproduced independent reference states and cell-type differences, revealed additional relationships and responses, and nominated candidate stress-response genes per cell type using Perturb-seq knock-outs.