Fig. 3

Prior/Intrinsic drug response associations. a Waterfall plot showing the number of agents with response that can be predicted (y axis; Fisher test BH p < 0.05) using knowledge of (prior/intrinsic) response of an agent (x axis). b Heatmap showing the odds ratios (log (OR))) of responding to one drug given knowledge of (intrinsic/prior) response of another drug, for all drugs in the panel with at least one significant predictive pair (BH p < 0.05). Red: positive association between drugs A and B (OR > 1; (intrinsic/prior) sensitivity to drug A predicts sensitivity to drug B, and (intrinsic/prior) resistance to A predicts resistance to B); Blue: negative association between drugs A and B (OR < 1; (intrinsic/prior) resistance to drug A predicts sensitivity to drug B, and vice versa). Two main clusters, cytotoxic vs. targeted agents, resolve into six main response clusters: (C1) DNA damaging/cell cycle, (C2) Aurora A kinase and cell cycle checkpoint inhibitors; (C3) microtubule poisons; (C4) AKT/mTOR/HDAC inhibitor mixed cluster; (C5) HER2/EGFR inhibitors; and (C6) PIK3C catalytic subunit inhibitors. Color bars represent manually annotated generalized modes of action. Clustering was performed using complete linkage and Pearson correlation distance