What Knowledge Can Be Extracted from Single-Cell Data? An Empirical Analysis of Donor-Robust Cell-Type Annotation
An empirical analysis of human pancreatic single-cell RNA-sequencing data demonstrates that for abundant, canonical cell types, cell-type annotation models generalize robustly across donors with only modest performance drops when evaluated via leave-one-donor-out compared to random splitting, underscoring the necessity of explicitly stating the population and technical scope when claiming knowledge extracted from single-cell data.