Extended Data Fig. 2m. For visualization and clustering, the data tables of the two libraries were concatenated, and the combined table was further reduced to retain only the most variable genes using the method outlined in Macosko et al49, yielding 1,102 genes. t–SNE projection of the data was performed using default parameters14. To determine the set of genes contributing to the separation of cell clusters, differential gene expression analysis (DEseq) based on negative binomial distribution50 was conducted to compare gene expression profiles in cells in each cluster versus those in the rest of the data set. Genes were ranked by smallest P values (expressed in terms of –log10) and the list of significantly over-represented genes with –log10 (P-value) < 10 of each cluster is provided as Supplementary Table 3. Patterns of expression for the top 25 genes in each cluster are shown in Extended Data Fig. 2e–l). A very small group of hCS-derived cells clustered with the GABAergic interneuron subdomain, and differential gene expression indicated that these cells expressed TBR1, RELN, PAX6 and CALB2.
