An Automatic Design Method Of Drug Molecules Based On The Fragment

Lead optimization is an important part of drug discovery.Lead optimization is widely used to improve the pharmacokinetic properties of molecules,reduce the toxic and side effects of molecules,and escape from patent protection.Traditional lead optimization progress relies heavily on the personal experience of medicinal chemists.Computer-aided drug design methods can provide valuable design ideas to increase molecular diversity.What’s more,the computational prediction of molecular properties and activity is useful for virtual screening.In this thesis,we have developed a lead optimization tool based on fragment-based drug discovery methods.Two kinds of approaches were adopted in our current thesis.One is fragment replacement.Fragment replacement includes scaffold hopping and side-chain replacement.The purpose of fragment replacement is to improve the pharmacokinetic properties of the molecule while maintaining the protein-ligand binding mode.The other is fragment growing.Protein three-dimensional structure is needed for fragment growing.The program works by finding suitable fragments for a protein pocket.Selecting a suitable fragment is the core of lead optimization.To construct a fragment database,we cut small molecules into fragments on all acyclic single bonds.Then,we calculate different kinds of descriptors.These descriptors were used for fast database searching.When users want to replace part of the molecule,the program searches the fragment database and assembles new molecules automatically.Finally,the program evaluates the newly designed ligands through the prediction of binding affinity and synthetic accessibility.Our program usually outputs the top two hundred analogs for further analysis.Our web server also provides detailed three-dimensional protein-ligand interactions for the designed molecules.The program developed in the current thesis is useful for fast fragment replacing or fragment growing which can assist medicinal chemists in lead optimization.We provide a web server for rational drug design.The web server is user friendly to none computational experts.