Extended Data Fig. 1 related to Fig. 1. Isolation and characterization of ES cell derived neural progenitor cells.

a. Schematic of our differentiation model including the specific days of sample collection. Human ES cells were differentiated into neuroepithelial (NE) cells using dual inhibition of TGFb and BMP followed by the transition to neural base media. Subsequently, sonic hedgehog and FGF8, are used to transition to the early radial glial stage (ERG). For the rest of the differentiation experiment the cells were constantly maintained in FGF2 and EGF2 neural base media to reach the mid radial glia (MRG) stage after 35 days, the late radial glia (LRG) stage after 80 and the long term neural progenitor (LNP) stage after about 200 days of in vitro culture. Cell type names indicated in red were profiled for gene expression, histone modifications as well as DNAme by WGBS, while names shown in grey for gene expression only and names in black for DNAme by RRBS only.

b. Hierarchical clustering for all RNA-Seq datasets collapsing replicates using the Jensen-Shannon divergence as metric.

c. Gene expression patterns shown as z-scores for all differentially expressed genes (q-value≤ 0.1) across ES cells and four neural precursor differentiation stages for genes expressed with at ≥ 2 FPKM in at least one stage (n=20,306). Genes were grouped into 18 clusters based on minimal average silhouette width using PAM clustering and Jensen-Shannon divergence based metric. Pie charts below indicate fraction of up (red) and down-regulated (green) genes during each transition.

e. Gene expression patterns shown as z-scores for all significantly differentially expressed genes (q-value≤ 0.1) across four more mature cell populations obtained through differentiation of NE, ERG or MRG cells to neuronal like cells (NE/ERG/MRGdN) and astrocyte like cells (LRGdA) derived from the LRG stage. Genes were grouped into 12 clusters based on minimal average silhouette width using PAM clustering and Jensen-Shannon divergence based metric.