Fig. 2: Pan-cancer summary of SLMs and validation in an isogenic screen.
From: The transcriptomic architecture of common cancers reflects synthetic lethal interactions
a, Bar plot showing the number of candidate synthetic lethal genes identified from CRISPR–Cas9 screens and the number of hyperexpressed synthetic lethal genes associated with specific TSG loss from TCGA tumor transcriptomes. b, Summary of observed SLM sizes expressed as empirical cumulative density function. Cumulative probabilty of observing SLM size ≤5, ≤25 and ≤50 are highlighted with dashed gray lines. c, Box plots illustrating hyperexpression (effect size >0) in RB1-defective patients in different TCGA cancer types. Dots indicate genes in cancer-specific SLMRB1 (hyperexpression effect size ≥0.25, FDR-adjusted P < 0.05 and synthetic lethal effect size ≤−0.25 and P < 0.01, with P values calculated using unpaired, two-sided, moderated Student’s t-test). The size of the dot shows the magnitude of the synthetic lethal effect size. Example RB1 synthetic lethal effects are highlighted. The number of genes in SLMs are shown in parentheses. Box plots show the first quartile (Q1), median and third quartile (Q3), with whiskers extended to points closest to Q1 − 1.5× the IQR (interquartile range) and Q3 + 1.5× the IQR. d, Box plots illustrating hyperexpressed genes in CDH1-defective patients in different TCGA cancer types. Dots indicate genes in cancer-specific SLMCDH1. Example CDH1 synthetic lethal effects are highlighted. Statistical thresholds are the same as in c. e, Box plots illustrating hyperexpressed genes in PTEN-defective patients in different TCGA cancer types. Dots indicate genes in cancer-specific SLMPTEN. Example PTEN synthetic lethal effects are highlighted. The statistical thresholds are the same as in c. f, Scatter plot illustrating RB1 synthetic lethal effects identified using an isogenic RB1 CRISPR–Cas9 screen in MCF10ATP53−/− cells. Delta (Δ) represents the difference in viability (normalized z-score) between MCF10AIsogenic and MCF10AParental cells. Key RB1 synthetic lethal genes are highlighted. g, Top, Venn diagram showing the overlap between DepMap-identified RB1 synthetic lethal genes and RB1 synthetic lethal genes identified using the MCF10A isogenic RB1 CRISPR–Cas9 screens. Bottom, Venn diagrams showing the overlap between RB1 synthetic lethal genes found in both DepMap and isogenic RB1 screens (74 genes) and hyperexpressed genes in RB1-defective cancers from SCAN-B and TCGA TNBC datasets. Over-representation P values are reported for all Venn diagrams (Fisher’s exact test).
