Extended Data Fig. 9: Variant-by-variant detailed representation of all analyses for TP53 variants.
From: Massively parallel phenotyping of coding variants in cancer with Perturb-seq
a. Variant features. Number of cells (y axis, top), distribution of normalized variant barcode expression (y axis, middle; red: variants with a fold-change greater than 1.5) and sc-eVIP scores (y axis, bottom; black: significant scores) for each variant (x axis), ordered as in Fig. 2d. Gray font: controls (synonymous and ExAC), blue font: hotspot variants (positions 175, 248, 273). b. Agreement with other data features. Top: difference (dot color) in mean expression or signature score between a variant (columns, ordered as in Fig. 2d) and unassigned cells and the significance of this difference (dot size, -log10(adj. p-value, Benjamini-Hochberg procedure), Kolmogorov-Smirnov two-sided test, Methods) for each of two genes canonically induced by TP53 and two TP53-associated signatures (rows). Colored border: BH FDR<10%. Middle: Growth (z-score, color bar) in three functional assays (rows) of each variant (columns). Bottom: Mutation prevalence (log2(counts+1) of variant occurrences) in two datasets (rows) of each variant, ordered as in Fig. 2d. c. Gene programs association with variants. Top: Difference (dot color) in mean program score (top) or mean PC score (bottom) between a variant (columns) and WT overexpressing cells and the significance of this difference (dot size, -log10(adj.p-value, Benjamini-Hochberg procedure), Kolmogorov-Smirnov two-sided test, Methods) for each gene program (top, rows, Methods), or each of the top 10 PCs (bottom, rows). Colored border: BH FDR<10%. d,e. Relation of variants to different clusters and cell cycle phases. Left: Proportion of cells (bar height) in each cell cluster (d) or cell cycle phase (e) (rows) derived for each variant (columns), annotated at the top with significance from a chi-square test comparing the cell state distribution of each variant with that of WT overexpressing cells (-log10(adj. p-value, Benjamini-Hochberg procedure)). Right: UMAP embedding of single-cell profiles, colored by cell clusters (d) or cell cycle phase (e). f. Relation of variants to the TP53 protein structure. sc-eVIP scores (y axis) of each variant (dot, colored by the variant class) and its position along the TP53 gene (x axis, annotated by domain). g. Variant induced shift in cell distributions. Density map of cell profiles in a UMAP embedding, comparing the density of cells overexpressing variants in each of 3 classes to either cells with variants in the WT-like class (grey, top) or unassigned cells (purple, bottom).
