Fig. 4: Prospective study design and random validation results.
From: A Bayesian active learning platform for scalable combination drug screens
a Prospective study cell line library broken down by type. b Prospective study drug library broken down by mechanism of action. c Overview of prospective study: after 15 rounds of BATCHIE data collection, approximately 4% of possible combinations were observed. Created in BioRender. Tansey, W. (2024) BioRender.com/a40a996. d Observation breakdown by cell line and drug mechanism of action; p-values were computed using Pearson’s chi-square test where the null hypothesis is that experiments were sampled uniformly at random. Colors in the left panel match the corresponding type colors from (a). e Scatter plot of mean BATCHIE predictions v.s. observed viabilities on random validation data. Orange line indicates regression of predictions onto observations, and black line denotes the identity line. f Pearson correlation between predictions and observed viabilities, broken down by cell line, observation status, concentration and cancer type. Maximum p-value satisfies p < 10−15, with no corrections made for multiple comparisons. g ROC curve for synergy identification on random validation data. Synergy is defined here as having an observed Bliss score larger than 0.25. p-value was computed using a one-sided permutation test over 100K permutations. e, f ρ is Pearson’s ρ correlation coefficient, with p-values calculated under a two-sided alternative. Source data are provided as a Source Data file.
