Our research is dedicated to the development and application of computational drug discovery methods to gain insight into the processes associated with protein-ligand and protein-protein binding, the accumulation of peptide-drugs during formulation, transport processes of ligands in proteins and proteins in cells, and pharmacokinetic properties and adverse effects of ligands.
We apply our computational techniques to design molecular probes or lead compounds that manipulate protein-ligand and protein-protein interactions. Current research activities focus on addressing serious shortcomings of present computational methods attempt to attain efficient modeling of protein-ligand and protein-protein association: protein flexibility and dynamics, solvation effects, a reliable quantification of binding affinities, and kinetic properties.
Our overall philosophy for method development is to pre-calculate important facets of protein-ligand binding only once per protein target by accurate but time-consuming methods. This information, for example the protein’s desolvation free energy, is incorporated into molecular modeling methods such as docking to accurately but efficiently predict protein-ligand association for a large set of ligands (e.g. in virtual screening or lead optimization). Following this strategy in our current research we extensively adapt modern deep-learning technologies such as convolutional neural networks to the modeling of protein-ligand binding.
Our group also participated in the development of the VirtualToxLab (Biographics Laboratory 3R), an in silico tool for predicting the toxic potential (endocrine and metabolic disruption, some aspects of carcinogenicity and cardiotoxicity) of existing and hypothetical compounds (drugs, chemicals, natural products) by simulating and quantifying their interactions towards a series of proteins suspected to trigger adverse effects using automated, flexible docking combined with multi-dimensional QSAR.