Design Of Algorithms And Programs For Drug Discovery And Drug Targeted Virtual Screening

As a new method or technology for drug discovery, virtual screening has been appreciated by many research institutions and big pharmaceutical companies. Moreover, virtual screening has been involved into the pipeline of drug discovery and development as a practical tool complemented with high throughput empirical screening. With the development of structural biology, structural genomics, functional genomics, and proteomics, more and more three dimensional structures of biologic molecule will be expressed, which will change profoundly the idea and strategy of drug discovery. A new blueprint of new drug discovery will finally come into being -from genome to drug, in which virtual screening would play a vital role. Molecular docking is a critical problem in the process of drug discovery and design in virtual screening, also it is an advanced research field all over the world. This dissertation mainly emphasizes on development of optimization methods and programs for molecular docking program. Our goals are developing and designing of molecular docking program with independent intellectual property, accelerating the R&D (Research and Development) and independent innovation of drugs in China.Above all, the author introduces and reviews the current docking methods in drug virtual screening, points out their deficiencies and limitations. In order to address these issues, the author has developed three programs for drug virtual screening, and has validated them using plenty of examples, furthermore, two of them have been put into application.By constructing molecular docking optimization model based on information entropy, we develop an improved multi-population genetic algorithm; using elitist strategy, competitive mechanism, and crossover among multi-population, to ensure the diversity of the populations. Information entropy is employed in the genetic algorithm, which will detect the space where the optimum falls in, to direct the further orientation of the optimization; at the same time, narrowed space is used as the convergence criterion which effectively controls the convergence of the algorithm, ensuring that the genetic evolution can converge rapidly and steadily. Based on this novel algorithm, a new fast flexible docking program, GAsDock, was developed. In comparison with other docking methods, GAsDock can bring better results both in docking accuracy and virtual screening efficiency, and acquire much higher speed than other programs. The results indicate that the computational time of GAsDock was approximately in linear scale with the number of the rotatable bonds of the compound, which promises GAsDock most suitable for large-scale compounds virtual screening. Lastly, some practical applications of GAsDock have been introduced.To solve the problem that most scoring functions can not satisfy for all the cases, the author provides a new drug virtual screening strategy, called MOOD, based on the Multi-objective Optimization Method (MOO). In this dissertation, the author constructs molecular docking MOO model. Two Multi-objective Optimization Methods, Deb's ε-MOEA and Evaluation Function Multi-objective Optimization...