Fig. 5: Drug target gene dependencies changed in ER-independent vs. ER-driven tumors.

a Workflow for applying elastic net models trained on CRISPR loss-of-function knockout (KO) screen data in cell lines to predict gene-level dependency using tumor gene expression profiles. b Distributions of predicted dependency scores for the four drug target genes ESR1, CDK4, CDK6, and CDK2 vs. integrative clusters, with sample sizes n = 57 (IC1), 127 (IC2), 97 (IC3), 107 (IC4), and 39 (IC5). A lower score indicates stronger dependency and greater sensitivity to gene knockout. Statistical significance was determined using two-sided Wilcoxon rank sum test. c Relative change in dependency scores for the four drug target genes in Post vs. Pre tumors. d Relative change in dependency scores for the four drug target genes vs. the mutation statuses of ESR1, RB1, and TP53. The y-axis shows the signed log10(p) values representing the statistical significance of the change in gene-level dependency scores between two subgroups, determined by two-sided Wilcoxon rank sum test. The dotted lines represent the significance cutoff (p < 0.05). MUT: mutated. WT: wild type. For all box-and-whisker plots, the box is bounded by the first and third quartile with a horizontal line at the median and whiskers extend to the maximum and minimum value. Source data are provided as a Source Data file.