Fig. 6: Validation of top osteosarcoma hits.
From: A Bayesian active learning platform for scalable combination drug screens
a Scatter plot of predicted viability against observed viability on OST hit validation data, colored by cell line. ρ is Pearson’s ρ correlation; p-value computed under two-sided alternative. b Observed single cell line TI scores broken down by prediction status, where highly sensitive/sensitive/resistant/highly resistant correspond to TI scores in the range (0.2,1.0]/(0.1,0.2]/[0.2,−0.1)/[−1.0,−0.2), respectively (n = 61, 33, 32, 24); p-values were computed using a two-sided Mann-Whitney U-Test with no adjustments made for multiple comparisons. Box plot center lines denote means, box limits denote standard deviations, and whiskers denote extremal values. c, d Z-scores of RNA expression and CRISPR knockout scores from DepMap grouped by gene and colored by cell line. e Observed single cell line TI scores; entries correspond to largest mean observed TI score over 4 pairwise concentration combinations. Predicted effect designations are identical to those in (b). f Interpolated TI values for the drug triplet alisertib, talazoparib, trametinib on SJSA-1 and U2OS when the total concentration is held fixed at 1 μM; optimal concentration is denoted by a star. g Per-drug concentrations that optimize single cell line TI score as a function of total concentration. TZP = talazoparib, TRM = trametinib, ALS = alisertib. Source data are provided as a Source Data file.
