Fig. 5: Prospective study validation of top hits.
From: A Bayesian active learning platform for scalable combination drug screens
a Pipeline for top hit selection and validation in the prospective study. The model fit on BATCHIE-collected data is used to simulate outcomes for drug combinations at the low (0.1 μM) concentration. Those simulated values are collected into confidence-rated predictions of TI values. The top hits are then selected for further in vitro experimentation, collecting observations over the full dose-response matrix. Created in BioRender. Tansey, W. (2024) BioRender.com/l79r044. b Observation status of selected top hits in BATCHIE-collected data. c Mean predictions and observed viabilities for selected top hits at 0.1 μM concentration. Pearson’s correlation ρ = 0.92 with p < 10−15 with a two-sided alternative. d Histogram of observed single model EWS TI values for combinations at 0.1 μM concentration in BATCHIE-collected data (gray, n = 659) and top predicted hits (orange, n = 100). Percentiles are drawn with respect to BATCHIE-collected data. e Observed AUCs with respect to the TI dose-response surface for selected top hits and not top hits. d, e p-values computed using a two-sided Mann-Whitney U-Test. Source data are provided as a Source Data file.
