Computer-Aided Drug Design Targeting Protein Phosphorylation

作为长期目标的一部分，开发和应用计算方法来帮助药物靶向设计

As part of the long-term goal to develop and apply computational methods to aid the design of drugs targeting protein kinases and related proteins, this research focuses on the development and application of the ensemble docking method, and on the study of drug-binding kinetics. Protein kinases continue to be the main targets for drug discovery in this research. The approval of about 60 inhibitors of protein kinases as drugs, mainly for treating cancer, has demonstrated protein kinases as important drug targets. As over 500 protein kinases are present in human and many mutants are driving diseases, many more drugs can be developed by targeting protein kinases. Specific Aim 1 continues to develop and apply the ensemble docking method to drug discovery. Aim 1a tests the hypothesis that scores, or their derivatives, from ensemble docking could predict whether lung cancer patients carrying disease-driving mutants of protein kinases are responsive to approved drugs. Aim 1b continues to validate the use of machine learning to improving ensemble docking. The validation will include all the proteins in the Directory of Useful Decoys-Enhanced developed for evaluating the performance of docking methods. Ensemble docking/machine learning models for these proteins will be made available to other scientists through the web server EDock-ML. Scientists can submit a compound to EDock-ML and receive the probability that the compound to be active. Aim 1c identifies new drug leads for the protein kinase c-MET with the aid of EDock-ML. Specific Aim 2 continues to test a combination of simulation methods for rapidly identifying compounds with therapeutically useful drug-binding kinetics, using more experimental data that are becoming available. It uses steered molecular dynamics (SMD) simulation for fast initial screening of chemical libraries, followed by evaluating the most promising subset by expensive but more rigorous methods, including the umbrella sampling technique, the Markov State Model, and the milestoning method. As it is still challenging to calculate absolute dissociation/association rate from molecular simulations, using several methods employing different approximations will help to draw robust and unbiased conclusions. After validation, the trajectories from the simulation will be used to decipher the molecular mechanisms of drug dissociation from protein kinases, including the examination of the generality of a two-step dissociation mechanism that has already been identified. Understanding the molecular mechanisms can give hint on the design of drugs with therapeutically useful drug-binding kinetics. The projects are designed to be performed by undergraduates. Senior scientists will work alongside the students often so that projects with higher impact can be included.