Biological pseudo-time. The single-cell transcriptome dataset of 161 blastomeres from six Day 2 and eight Day 3 embryos was analysed with the CellTree tool24. (a) Backbone tree. A backbone tree is built by computing a matrix of pairwise distances. This shows the hierarchical relationship between all blastomere transcriptome profiles and aligns the blastomeres in a pseudo-time line. The backbone trees of individual embryos are sorted according to the median position of their cells. Blastomeres from Day 2 embryos are coloured in blue, blastomeres from Day 3 embryos in red. (b) Identification of latent groups of genes (topics), which characterise the steps of the development. The number and distribution of topics along the pseudo-time line was obtained by using the Latent Dirichlet Allocation (LDA). To each topic, gene ontology terms were associated. Topic 1 = “translation”, “cell division”; Topic 2 = “translation”, “regulation of translational initiation”, “mRNA splicing, via spliceosome”, “cytoplasmic translation”, “rRNA processing”, “spliceosomal complex assembly”, “negative regulation of mRNA splicing, via splicing”, “cell division”, “regulation of alternative mRNA splicing, via spliceosome”; Topic 3 = “translation”; Topic 4 = “ATP synthesis coupled proton transport”; Topic 5 = “mitochondrial translational elongation”; Topic 6 = “organic hydroxyl compound transport”.