Fig. 1: Generation of a spatio-temporal molecular atlas of the germ layers of gastrula-stage mouse embryo.

a Spatial domain of cell populations in the epiblast/ectoderm, mesoderm, and endoderm of E6.5–E7.5 embryos, defined by the position-specific expression of zipcode gene transcripts. Geo-seq sampling positions: epiblast/ectoderm—A, anterior; L, left lateral; R, right lateral; L1/R1, left/right anterior lateral, L2/R2, left/right posterior lateral; M, mesoderm—MA, anterior mesoderm; MP, posterior mesoderm; E, endoderm—EA, anterior endoderm; EP, posterior endoderm. Number: descending series indicating positions in the proximal-distal axis. Germ layer domains: Epi: epiblast, Epi1, 2, 3: epiblast domain 1, 2, and 3; M: mesoderm, M1, M2: mesoderm domain 1 and 2; MEP, putative mesendoderm progenitors; E: endoderm, E1, E2, E3: endoderm domain 1, 2, and 3; PS, primitive streak. b The structure of the Population Tracing algorithm for imputing the developmental connectivity of cell populations across stages of gastrulation (see the detail of mathematical operations in Methods and Supplementary Fig. 2f). c The developmental trajectory of sub-populations within each germ layer tissue domain descending from blastomeres of the preimplantation E2.5 morula stage embryo to the germ layers of E7.5 late-gastrulation stage embryo. d 3D Model of the epiblast/ectoderm displaying the cell populations by imputed positional coordinates (see Methods for details of the mathematical modeling). The exemplar 3D corn plots show the spatiotemporal distribution of the Mixl1-expressing population in the primitive streak; the proximal-distal span of the Mixl1+ domain defines the developmental stage of the gastrulating embryo. The color legend indicates the level of expression determined by the transcript counts. e A flow diagram of the 4-step spatial mapping protocol. (1) Multi-Dimension Single-Cell (MDSC) Mapping allocates single cells to their imputed position. (2) The Annulus Model simulates the Geo-seq positions. Single cells that mapped to a Geo-seq position were distributed uniformly across the interior space of the position in each annulus section. (3) Bubble Sort algorithm displays the cells in relation to the gradient of gene expression level or signaling intensity. (4) The optimization algorithm refines and visualizes the spatial distribution pattern of cell types by optimal coordinates at each Geo-seq position.