Fig. 3: In silico perturbation of GBM tumor cell scRNAseq data using an L1000-derived alisertib TCS predicts an NPC-like to MES-like tumor response confirmed in vivo.

a Histogram of single-cell alisertib TCS connectivities. b Hierarchy plot of patient GBM tumor cells colored by predicted alisertib sensitivity (ρ < 0) or resistance (ρ > 0). c Box plot depicting per-patient difference in proportions of cells in each GBM cell transcriptional state in resistant vs. sensitive populations. Each dot represents the difference for an individual patient (n = 11). Boxplot elements represent the median (center line), quartiles (box limits), and 1.5x interquartile range (whiskers). Adjusted p-value from a two-sided Wilcoxon signed-rank test with Benjamini–Hochberg correction (MESvsNPC p.adj=0.019). d Schematic of in vivo experiments. (Created in BioRender. Suter, R. (2025) https://BioRender.com/3fhf7s6). e UMAP plot of pre-filter single-cell transcriptomes colored by treatment with alisertib or DMSO. f UMAP of captured cells colored by percent alignment to the human transcriptome (hg19). g Dot plot of pass-filter single-cell transcriptomes showing Neftel et al. signature expression, grouped by predominant transcriptional state module expression. h Two-dimensional hierarchical representation of GBM22 xenograft cells’ relative enrichment scores for transcriptional state modules. Cells are colored by assigned transcriptional state identity. i Violin and box plots quantifying enrichment shift of transcriptional state signatures in alisertib-treated xenograft cells (n = 10,394, cells from 3 pooled xenografts) normalized to DMSO-treated xenograft cell mean enrichments (n = 5708, cells from 3 pooled xenografts). Boxplot elements represent the median (center line), quartiles (box limits), and 1.5× interquartile range (whiskers). Violin width indicates the kernel density of the data. (Two-sided, one-sample Wilcoxon vs 0 with Benjamini–Hochberg correction: *p-adjusted=8e^-9; **p-adjusted=1.1e^-60;***p-adjusted=3.2e^-238; NS p-adjusted=0.79). j Alluvial plot depicting shift in relative proportion of transcriptional state identities in alisertib and DMSO vehicle control-treated xenografts. k Scatterplots of L1000 small molecules and subsets of compound classes’ TCSs depicting log₂FC predicted correlation shift vs. observed correlation shift in alisertib-treated xenografts. Differential small molecule correlations were calculated using limma. Spearman correlation coefficient R (ρ) and p values were determined using a two-sided Spearman’s correlation test (degrees of freedom=n-2). Shaded regions represent the 95% confidence interval of fitted regression lines. Source data are provided as a source data file.