Fig. 1

eSPRESSO analysis for graph-based SOM clustering to detect spatial discriminator genes (SDGs). a Schema of the algorithm. eSPRESSO performs SOM clustering using MCMC algorithm with gene expression data based on graph representation of topology among cell or tissue types. A, anterior; P, posterior; L1, anterior left lateral; R1, anterior right lateral; L2, posterior left lateral; R2, posterior right lateral; MA, anterior mesoderm; MP, posterior mesoderm; EA, anterior endoderm, EP, posterior endoderm; Ect1–3, ectoderm; PS, primitive streak; E1–3 (En), endoderm; rep.1–8, replica. b Optimization of SDGs by replica exchange to increase cell or tissue type clustering accuracy while preserving topological consistency defined by weighted ARI + accuracy score. c A topological distance map of cells or tissues for the original (lower left) and resultant (upper right) clusters. d Gene expression heatmap for the optimized SDGs. e 3D reconstruction of cell or tissue types with SDGs using UMAP