Theoretical studies of protein-ligand binding

Sammanfattning: Understanding how drugs work is of great importance, since it can facilitate drug discovery, both time- and costwise. At the same time, it is important to have methods that can help predict how well does a potential drug molecule bind to its target. Computational methods can in many ways contribute to drug design process. In this thesis, we employ different computational approaches to study the binding of various ligands to galectin-3 protein, which is an excellent model system and an interesting therapeutic target. We study the effects of solvationthermodynamics, protein and ligand conformational entropy, as well as specific protein–ligand interactions. Our results indicate that accurate modelling of protein–ligand binding requires careful consideration of solvation and protein ligand conformational entropy, since they contribute significantly to protein-ligand binding free energies. We also compared different methods used to study the water structure and thermodynamics in the protein–ligand binding site, where we showed that solvent-exposure of the binding site may dictate the choice of the method. Moreover, we participated in the D3R and SAMPL6 blind challenges, where we tested the performance of different methods used to estimate binding affinities. We have showed that the predictions of relative binding affinities improve if displaced water molecules are included in the free-energy perturbation calculations and if the ligand is treated by quantum mechanical methods.