Extended Data Fig. 10: Validation of GSC-state CNVs in patient tumors and identification of GSC-like tumor cells.
a, Genome-wide inferred CNV profiles for 14,207 malignant cells from 7 patient tumors. Columns represent genomic regions, ordered by genome position across all chromosomes. Rows represent CNVs for individual cells, annotated by sample. b, Developmental (left) and Injury Response (right) program scores across quartiles. Numbers underneath quartile labels depict the number of cells harbouring respective Developmental or Injury Response CNVs. Enrichment of CNVs between upper and lower quartiles was determined using a Chi-squared test. Box plots represent the median, first and third quartiles of the distribution and whiskers represent either 1.5-times interquartile range or most extreme value. Outliers are displayed as circles. c, Train and test accuracy for logistic regression model, 30 random 80:20 train test splits (left). Distributions of model coefficients corresponding to the 30 trained models (right). Model coefficients are weights by which the logistic regression model describes class likelihood as a function of PC1 and PC2. Box plots represent the median, first and third quartiles of the distribution and whiskers represent either 1.5-times interquartile range or most extreme value. Outliers displayed as circles. d, Proportion of cells in GSCs correctly classified as being GSCs (blue) or misclassified representing tumor-like GSCs (white). Proportion of tumor cells correctly classified as being tumor (black) or misclassified as being GSC-like (grey). e,f, PCA plot of all GSCs and tumor cells as in Fig. 5a. Black line represents contour encompassing 99% of tumor cells. Blue line represents contour encompassing 99% of GSCs. Grey dots represent tumor cells classified as being GSC-like. White dots with blue outline represent GSC cells classified as being tumor-like. g, Differential gene expression analysis between tumor cells and GSC-like tumor cells. Each dot represents a gene (x-axis) ordered by average log2 fold change (y-axis). Red dashed line represents a log2 fold change of double between groups. h, Differential gene expression analysis between GSCs and tumor-like tumor cells. Each dot represents a gene (x-axis) ordered by average log2 fold change (y-axis). Red dashed line represents a log2 fold change of double between groups. i, Expression of mature and young astrocyte gene signatures between tumor cells (black; n = 12,145 cells) and GSC-like tumor cells (grey; n = 2,062 cells). j, Expression of mature and young astrocyte gene signatures between GSCs (blue; n = 64,502 cells) and tumor-like GSCs (white; n = 1,153 cells).
