Study On The Computational GPCR–Ligand Docking

G-protein coupled receptors (GPCRs), a kind of very important signal molecularreceptors, play important roles in life process, and are the targets of about 50% moderndrugs. With the computing power increased dramatically in the past decade andthe booming of the knowledge about the interaction between physiology molecularand biology molecular, it becomes possible for the prediction of accurate protein–ligand binding mode in a computational way. The study of computational GPCR–Ligand docking methods is important to the computer aided drug research and otherapplications.Computational docking methods usually generate lots of GPCR–Ligand confor-mation decoys. How to identify the near-native conformation from decoys is a hardproblem. The thesis presents a SVR-based method to identify near-native confor-mations from decoys by predicting the ligand RMSD of the complexes. The SVR usevarious energy score items and other ligand position-related information as the featuresfor training the predictor.3D homology models are still the main sources of GPCRs rational drug design,thus it is important to support receptor structure flexibility while docking. Twoschemes for docking with flexibility are proposed in terms of local and global flexi-bility. The first is to introduce the refinement of the receptor's local structure duringthe docking process. The flexible region is recognized according to the current dockingposition of the ligand, and then Rosetta loop refinement protocol is used to simulatethe flexibility. 5 GPCR homology receptors are docked to a ligand with this method,and the results show that the scheme rescues the docking for those homology recep-tors with low accuracy. The second scheme is to integrate receptor conformers whiledocking with a ligand. In such a way, global flexibility of the receptors is employed tothe docking procedure.The SVR method to identify the near-native conformation from the docking re-sults releases the pressure for docking to generate high-accuracy GPCR–Ligand com-plexes. On the other hand, introducing the receptor flexibility in docking is helpfulfor generating high-accuracy decoys. This thesis contributes to the exploration of GPCR–Ligand docking with high accuracy.