Extended Data Fig. 8: Performance of CONCORD on public human and mouse scRNA-seq datasets.
From: Revealing a coherent cell-state landscape across single-cell datasets with CONCORD
UMAP embeddings derived from each method’s latent space are shown for all datasets and colored by batch and cell type. a, Lung atlas spanning multiple spatial regions, donors, and two scRNA-seq protocols34. b, GTEX v9: human single-nucleus RNA-seq data from eight tissue types across 16 individuals80. c, HypoMap: single-cell atlas of the murine hypothalamus ( ~ 380k cells) across four assays81. d, Immune cell atlas: human immune cells from 16 tissues and 12 donors82. e, Mouse pancreatic islet: scRNA-seq atlas comprising 56 samples across sex, age, and diabetes models83. f, Tabula Sapiens: human cell atlas of over 1.1 M cells from 28 organs of 24 normal human donors66. Missing plots indicate runs that failed due to excessive resource demands or violated model assumptions.
