| Molecular docking is a powerful approach in computer-aided drug discovery and drug design. The computer mode identification and optimization technique are used to search the special molecule in three dimensional small molecular database, the special molecule matches with the active site of the specially designed target relating disease in geometry, physics and chemical characteristics. In this way, the computer screening (virtual screening) can be used to drug design. Virtual screening has not only speed the process of new drug research and development, but also reduces products costs gristly. Molecular docking technique is the crucial problem in virtual screening, and at the same time is the forward problem of drug design in the world. The research work of the dissertation expends along above direction.First of all, a summary of the status of molecular docking in drug design is introduced. Based on further studying for the molecular docking theory, an entropy based multi-population parallel evolutionary model of the molecular docking in drug design is constructed, and an improved algorithm and parallel computing strategy based on model level is studied, the parallel grain is increased by means of independent genetic operating at population level, and "inbreeding" problem within one population is avoided by crossing among different small populations. On the premise of keeping the diversity of members of populations, it reduces the size of every population in traditional multi-population genetic algorithm and optimizes new generation. The equivalence on the optimal solution between the new model with information entropy optimization and original model of the optimal design for molecular docking is proved in this dissertation. The introducing of information entropy makes the implied evolutionary aim enhanced greatly, and the deficiency of time consumption and slow convergence speed are overcome. Examples show that the method has good accuracy and gives very high speed-up and efficiency.The above method is integrated into the widely used docking software DOCK5.0 as the conformation search algorithm, and a new microcomputer version of docking software is developed. Further more, the multi-population evolutionary algorithm is paralleled with coarse grain under the MPI programming environment. The improved molecular docking software runs on the high performance computer Tianchao Dawning 3000. Comparing to DOCK5.0, it has higher computing efficiency.With the proposed software, the docking design of ligand and receptor in the crytal complex of COX-2 and its inhibitor, and the peroxisome proliferators-activated receptor r (PPARy) and its candidate agonist BRL49653 are given, and the results show that the optimized conformation is very close to the one of template molecule separately.Additionally, a small database composed of the 49 molecules supported by DOCK4.01 and the known biological active molecule celecoxib is investigated to test the efficiency of the proposed method and software in virtual screening of drug molecules. Results show that the developed software does screen out the celecoxib with the lowest binding energy than all the others, and its binding energy is better that of the template molecule. The program system is used to select the immunoadsorption materials and gives good results.This dissertation is financially supported by the National Natural Science Foundation(10272030) and Subsidized by the Special Funds for Major State Basic Research Project (G1999032805) of China.
|
PDF Full Text Request