Fig. 5: Cerebral gliomas show developmental identities that reflect a foetal-like nature.

a Left: deconvolution analysis heatmap of tumour clusters from 10 IDG patients from Venteicher et al., 2017 showing relative fractions of mouse cell-types for each patient cluster. Bars on the left indicate the state of the cells (malignant or non-malignant) and the type of glioma. Each patient sample was clustered separately (see patient clustering in Supplementary Fig. 7) and the clusters are denoted as e.g., “Patient#1-cluster#1” = “Patient 1-C1” (see Supplementary Data 1). Shown on the right is a bar plot summarizing the matching results and overlap size between the patient clusters and a schematic summary of the tumour lineage from the deconvolution analysis. b Deconvolution analysis heatmap of tumour clusters from 6 ODG patients from Tirosh et al. 2016 showing relative fraction of the mouse cell-types for each tumour cluster. The patient samples were analysed and clustered together (see Methods and clustering analysis in Supplementary Fig. 9b). c Deconvolution analysis of scRNA-seq tumour clusters obtained from four paediatric supratentorial ependymomas showing relative fraction of the mouse cell-types for each tumour cluster. The patient samples were clustered together and immune cells were identified and excluded using the expression of known immune markers (see Methods and clustering analysis in Supplementary Fig. 7d, e). Abbreviations as in in Fig. 4.