Figure 2.

Learned classification tree from independently annotated datasets

(A) UpSet plot visualizing the intersection and the number of common cell type annotations between the different datasets. Disconnected dots correspond to the number of unique cell types in each dataset, while connected dots represent the intersection between the datasets. Additionally, the number of cell types identified in each study is plotted alongside each dataset.

(B) UMAP representation of the originally annotated cell types across all datasets.

(C) Learned classification tree applying a k-Nearest Neighbor (kNN) classifier on the six annotated datasets. The color(s) of the tree nodes correspond to the supporting dataset(s) Missing populations: PTS1_Janosevic21, PTS2_Janosevic21. Arrows mark both inaccurate placements in the tree (Fib, Stroma) and cell types that can be further annotated to increase resolution IC: Intercalated Cell, ICA: Intercalated Cell Type A, ICB: Intercalated Cell Type B, Endo: Endothelial Cell, Fib: Fibroblast, Macro: Macrophage, B lymph: B lymphocyte, Stroma: Stroma cell, NK: Natural Killer, T lymph: T lymphocyte, PT: Proximal Tubule, PTS1: Proximal Tubule Segment 1, PTS2: Proximal Tubule Segment 2, PTS3: Proximal Tubule Segment 3, PTS3T2: Proximal Tubule Segment 3 Type 2, PC: Principal Cell, PEC: Parietal Epithelial Cell, Per: Pericyte, DCT: Distal Convoluted Tubule, ATL: Ascending Thin Limb of Henle, MD: Macula Densa, LOH: Loop of Henle, CTAL: Thick Ascending Limb of Henle in Cortex, MTAL: Thick Ascending Limb of Henle in Medulla, CNT: Connecting Tubule, Podo: Podocyte, DTL: Descending Thin Limb of Henle, MC: Mesangial Cell, Neutro: Neutrophil, Asc-Vas-Recta: Ascending Vasa Recta, Desc-Vas-Recta: Descending Vasa Recta, Glom Endo: Glomeruli Endothelial.