Fig. 4

Identification of known genotype-specific essentiality. A Estimates of essentiality per gene (\(\widehat \phi \); x-axis) and essentiality log likelihood ratios (Ess. LLR; y-axis) from ACE based on genome-wide CRISPR-screen data from the Achilles DepMap Project. Results are shown for 688 “essential” genes (Pos. Controls; red) identified in references [1–3] and 634 nonessential genes (Neg. Controls; blue; see the “Methods” section for details). The LLR metrics are relative to a null hypothesis of \(\widehat \phi =0\). B Estimated essentiality per gene in wild-type TP53 (ϕ; x-axis) vs. log likelihood ratio (LLR) for differential essentiality in wild-type vs. mutant TP53 based on DepMap data for NSCLC Adenocarcinoma [9]. Several high-scoring genes are highlighted with labels, including MDM2 (see text). C Top 10 genes identified as differentially essential in NSCLC samples having wildtype TP53, ordered by differential essentiality log likelihood ratio (Diff. Ess. LLR). D Top 10 genes identified as differentially essential in NSCLC samples having mutant KRAS. In panels A–D, the horizontal dashed lines indicate Bonferroni-corrected empirical p-values of 0.05. Genes highlighted in red in panels C and D are discussed in the text. E Ranking of positive (CDKN1A) and negative (MDM2) regulators of TP53 by differential essentiality test statistic across all prediction methods. Ranking is relative to 1171 uniformly essential or nonessential genes from our previously described negative and positive control sets; a subset of 50 from each of these sets were provided to methods as annotated controls. An illustration of MDM2 alone is shown in Additional file 1: Fig. S9. F Median ranking of 19 differentially expressed genes and their 14 associated genetic backgrounds across five tissue types from DepMap’s Project Achilles. Ranking is relative to 500 uniformly essential and 500 uniformly nonessential genes from our previously described negative and positive control sets. Method performance by gene and tissue type is shown in Additional file 1: Figs. S11 and S12. Results from a one-sided test using expected direction of essentiality are shown in Additional file 1: Fig. S13