Extended Data Fig. 2: Integrated single-cell transcriptomics analysis of human embryos and 3D-hE-gastruloids.

a, t-SNE embedding of 394 single-cell transcriptomes showing the results of clustering analysis for the identification of cell populations in the 3D epiblast model dataset. Cells are coloured by their differentiation day (left) and inferred cell type (right). b, Barplots showing the relative percentage of cells by differentiation day in each cluster/inferred cell population. c, Box and dot plots showing the expression levels by cell type of selected marker genes for pluripotency, epithelial, polarity, primitive streak and amnion cell identities (total number of cells = 394). Expression is reported as log-normalized counts. Horizontal line indicates median, box indicates the interquartile range (IQR) and whiskers denote the 1.5 × IQR. Coloured dots indicate the cell cluster in which the gene was found as a marker. d,e,f, t-SNE embedding of 892 single-cell transcriptomes coloured according to the relative (Z-score) expression levels of selected marker genes for primitive streak (CDH2, EOMES, VIM), ICM (ESRRB), Pre-implantation epiblast (KLF17, TFCP2L1), general pluripotency (PRDM14, SOX2) and epithelial/polarity (CLDN6) cell identities. g, UMAP embedding of 337 single-cell transcriptomes showing the results of pseudo-temporal ordering by reverse graph embedding using Monocle3. The line plot on the leftmost UMAP represents the embedded trajectory graph. Cells are coloured according to pseudotime (left), differentiation day (middle) and inferred cell type (right).