Fig. 2.

Individual model fit examples. The estimated parameters of the GPDI model were used for simulation (red) and compared to observed time-courses (black) exemplified in a–d for 1 out of the analyzed 200 combination scenarios; the combination of Bromopyruvate (Bro, drug A) and Staurosporine (Sta, drug B) (a–d), at concentrations AxBx (relative minimal inhibitory concentration²³); The Loewe Additivity (a), and Bliss Independence-based (b), GPDI models described the 64 experiments of this combination scenario very well, the monodirectional nature of this interaction was indicated by the INT parameter values (fractional change of victim EC50) at the perpetrator EC50 that identified Bro as sole perpetrator drug in this combination; the empiric Loewe Additivity-based Greco (c), and empiric Bliss Independence (d), model solely quantified antagonism, but did not describe the data well, with prevalent with over- and underprediction; Predicted vs. observed fungal load for all 1.23 million observations from all 200 combination scenarios indicated superior predictive performance of the GPDI models (e, f), compared to the Greco and empiric Bliss model (g, h)