Molecular Docking Conformation Search Optimization Strategy And Algorithm Research

In the 21 st century, new drug research focused on the excavation and design of lead compounds, which CADD is one of the important methods of the design of lead compounds. Due to the outstanding performance and highly efficient of finding lead compounds, molecular docking has been widely used in drug researching. When docking, firstly put the ligand on the active sites of the receptor, in this process, a lot of optional matching patterns are required. According to the energy complementary, geometry complementary, chemical environment complementary principles, identify the most stable mode which got the lowest score of energy. Determine the binding strength between molecules and the optimal binding position of docking are the issue of docking faced, meanwhile, the evaluation function and conformation search is the molecular docking study two important areas. As the ligand and receptor molecules can freely rotate and move and the conformation of the two molecules can changes, this makes the possibility of a large number of possible patterns, simple search method cannot accomplish this daunting task, high-optimization method have to be introduced.Genetic algorithm and its derivatives are the common way to search conformation. Optimization conformational search algorithm and its model is the main content of this paper, the classic genetic algorithm is the foundation. After introducing the molecular docking related knowledge and the docking principles, elaborating the classic genetic algorithm in detail, this paper focus on the study of conformation search algorithm and proposed a docking model based on Lamarck genetic algorithms. The lack of classical genetic algorithm in terms of local search can be marked up by simulating the individual in the life cycle. Information entropy, elitist strategy and multi-population are included in this model, so that can accelerate the evolution while ensuring the diversity of the population. However, the lack of Lamarckian learning is not effectively distinguished between individuals, resulting in a waste of local research resource. This paper then improved the model proposed, in order to tap the potential, the individuals processing Baldwin learning before Lamarck learning, this can make Lamarckian learning be more targeted. Learning potential and fitness of individuals are the evaluation. The commonly used docking software includes DOCK, AUTODOCK and FLEXX, these software have advantages and disadvantages. AUTODOCK4.2 is open source software. This paper based on AUTODOCK4.2 source code and applied the model proposed before on it. Compared with the original experimental results, validate the model presented in this paper can really effectively applied molecular docking.