Extended Data Fig. 8: Comparison of ESC in vitro differentiation protocols to in vivo populations.

(a) UMAP dimensional embedding of 2D-ESC and 2D-ESC-Pi3ki datasets. (b-c) Sankey diagrams representing significant alignments between clusters from 2D-ESC + PI3Ki protocols and our in vivo dataset. The alignments to DE2 and hindgut lineages are highlighted in red in (b) for fraction of 2D-ESC cells and in (c) for fraction of 2D-ESC-PI3Ki cells. The height of the cluster-bars to the left represents the proportion of cells in each cluster from (b) 2D-protocol and (c) PI3Ki-protocol. The width of the lines is given by the cluster-height divided by the number of alignments. D4 – Day 4, D5 – Day 5, D6 – Day 6. (d) UMAP dimensional embedding of 2D-ESC and 3D-ESC protocol highlighting that even when 2 different cell lines are used, the 2D and 3D protocols cluster together. (e) Scheme of calculating similarity measure S = 1-n_DE/N_E (1-n_EE/(n_DE (n_DE-1))). In this example, the total number of in vivo lineages, N_E = 5; in the bottom right panel, number of alignments, n_DE = 4 (red and green arrows); number of alignments among the four in vivo lineages n_EE = 2. DE1- early definitive endoderm, DE – definitive endoderm, DE2 – later definitive endoderm, FP – floorplate progenitors, PS2 – later primitive streak.