The expression of the tissue-specific genes in vulnerable tissues is consistent with a PE mechanism and the effect of copy number variations (CNVs). a) The average standardized expression of tissue-specific genes in vulnerable and nonvulnerable tissues. b) The observed mean absolute t-statistic (dashed red line) was calculated for differences in the expression of tissue-specific genes between the vulnerable and nonvulnerable tissues. The histogram shows the distribution of the expected values under the null based on 10,000 permutations. c, d) Demonstration of PE. c) MYB, a tissue-specific gene in blood and lymphocytes, is more essential (lower dependency) in cells with higher expression of MYB. d) SOX10, a tissue-specific gene in the skin, is more essential in cells with higher expression of SOX10. e–g) Demonstration of a tissue-specific gene that is explained by the gene copy number. e) BUD13, a tissue-specific gene in the skin and the upper aerodigestive tissue, is less essential in cells with higher expression of BUD13. f) BUD13 expression is associated with the copy number of BUD13. g) BUD13 dependency scores are associated with the copy number of BUD13. h) Density plots of Spearman’s correlation coefficients between the dependency scores and expression of tissue-specific genes. The plots are shown for 623 tissue-specific genes with CNV information, divided into genes with significant correlation with the gene copy number (FDR < 0.05) (light red) or nonsignificant (grey). All P-values presented in the figures are adjusted for multiple tests by the Benjamini and Hochberg FDR procedure. The yellow dots (C-G plots) represent cell lines that belong to the vulnerable tissues, and the blue dots are for cell lines in the nonvulnerable tissues. In C-G, locally estimated scatterplot smoothing (LOESS) curves and 95% confidence intervals are shown alongside Spearman’s ρ and the FDR value.