Fig. 4

Sorted, fixed mouse brain cells allow identification of distinct neural and non-neural cell types. Hindbrains and cerebellum from newborn mice were microdissected and dissociated, and cells were sorted by FACS into methanol and stored. Drop-seq data correspond to two independent biological replicates. Libraries were sequenced to a median depth of ~7100 reads per cell. a Distribution and the median of the number of genes detected per cell (>300 UMIs) in Drop-seq data pooled from two Drop-seq runs, representing 4366 cells. Note that violin plots are displayed on a log scale. b Clustering of 4366 fixed cells into distinct cell populations marked by colour (Additional file 8: Table S2). The plot shows a two-dimensional representation (tSNE) of global gene expression relationships among all cells. Tissue associations of cell clusters were identified by assessing the 50 most variable genes in each cluster and confirmed by inspection of publicly accessible images of RNA in situ hybridizations. c Known marker gene expression in clusters of brain cells (see text for explanation). Expression coloured based on normalized expression levels