Authors

M De Rycker¹;  ¹ University of Dundee, UK

Discussion

The development of new treatments for parasitic diseases is a complex multifactorial process, with high rates of attrition. Computational approaches are increasingly applied to expedite the development of new treatments. Methods to interpret large datasets, build predictive models and design new compounds all contribute to better decision making. Here I will outline a series of computational approaches that we are using or developing for our kinetoplastid and apicomplexan drug discovery programmes. We are applying data-driven deep learning models to predict activity against key undesirable targets for Trypanosoma cruzi, allowing us to screen bespoke library collections depleted of likely non-progressable compounds. Using high-content imaging and image-based profiling we are developing methods to cluster compounds by their resulting phenotype and the underlying mechanism of action. Finally, I will introduce quantum mechanics approaches to understand ligand–protein interactions driving the molecular recognition process and to build predictive affinity models to guide compound design.