In this contribution, we will summarize five studies where we have recently modelled compound combination effects. These studies are in the areas of large-scale cancer combination screens, interactions between antibacterials, synergy in antimalarial assays, modelling a Traditional Chinese Medicine, and understanding the mode of action of a complex molecular mixture.<br/>Our first study was concerned with the modelling of a large-scale cancer cell line screen, comprising 14 experimental and 24 approved drugs screened in 39 cancer cell lines. At the chosen binary classification cut-off, we were able to achieve a true positive rate of 0.55 at a true negative rate of 0.95, representing the ability to reject 19 out of 20 non-synergistic combinations while one out of every two positive predictions is a true-positive. In the second study, we have attempted to predict synergistic antibiotic interactions among 18 antibacterials against <i>E. coli </i>using descriptors derived from chemical structure. It was found that chemical structure is a surprisingly good descriptor of synergy, more so than predicted protein targets of compounds, and that the applicability domain is an important consideration when predicting compound synergies. In the third study we explored using gene expression in combination with HTS data for selecting synergistic compound combinations, and some of the selected compounds were prospectively shown to be synergistic across three antimalarial strains. In the area of Traditional Chinese Medicine (TCM), we next explored to what extent network-based measures can be used to predict and model synergy of pairs of 22 compounds of Shexiang Baoxin Pill (SBP), and using network information we could make some predictions about synergy - about 70% of the 20 highest-ranked compound combinations according to predicted synergy were prospectively found to be synergistic, while this was only true for about 30% for the 20 lowest-ranked compound combinations. Finally, we utilized both chemical and gene expression information to understand the mode of action of a complex compound mixture, Abo1, which is currently under investigation as a potential mesothelioma treatment.<br/>To conclude, understanding drug interactions and modelling synergy are areas where much further research is needed, on the fundamental level – with respect to the systems we measure and define synergy in, and how such synergy is defined – as well as on the level of predictions and deriving biological insight.

Understanding drug and compound combinations and modelling synergy – Methods and applications

256th National Meeting (2018)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.