Fig. 3: Comparative analyses revealed human/macaque-biased CREs in glutamatergic neurons.

a UMAP co-embedding of snATAC-seq data from macaques, humans and mice. Colored dots at the bottom indicate human/macaque-biased glutamatergic neurons derived from snATAC-seq data. b Heatmap showing the similarity of snATAC-seq derived glutamatergic cell clusters across macaques, humans and mice using MetaNeighbor. c The outer box represents the number of CREs showing higher chromatin accessibility in human/macaque-biased than other cell types of the macaque cortex. The inner box represents CREs with high chromatin accessibility in the corresponding human cell types. The inner circle indicates CREs showing no homologous sequence in the mouse genome. d GO enrichment for genes linked to human/macaque-biased CREs (inner circle in c). e Genome browser tracks showing the chromatin accessibility of human/macaque-biased CREs located on the DCC gene in the human/macaque-biased and other cell subclasses. f Spatial map showing the chromatin accessibility of the human/macaque-biased CREs shown in e in a representative Stereo-seq section at indicated coordinate. The right panel showing the quantitative chromatin accessibility level across six cortical layers. The sample sizes are listed in Supplementary Data 10. Error bars represent SEM. g RNAscope results showing the layer 4-specific expression of DCC gene in TPO (left) and V4t (right) regions of the macaque cortex. Probe against DCC (ACD, cat#1843701-C1) was listed in Supplementary Data 9. DAPI was used for nuclear counterstaining. One of two biological replicates was shown from a single experiment. Independent replicates showed consistent patterns. h Scatterplots showing the expression of layer 4 marker gene RORB and DCC gene in human MTG of the previously published MERFISH dataset³⁴. i Bar plots showing 7 categories of CREs in various cell subclasses from macaques, humans and mice. j Boxplots showing the percentage of CREs colocalized with TE in different categories (N = 20 subclasses). P value was calculated using a two-sided Wilcoxon rank-sum test, adjusted by Benjamini-Hochberg FDR correction. k Boxplots showing the percentage of CREs located in different types of TEs (N = 20 subclasses). Statistical testing was performed analogous to j.