Research Of Grid Techniques And Application Related In Drug Grid

With the rapid development of X-ray crystallographic and nuclear magnetic resonance (NMR) techniques, more and more three dimensions (3-D) protein structures are known by people, which makes a promising future for DMD (drug molecular discovery). However, DMD suffers the problems that, firstly, the searching space of drug molecular docking is huge, which includes 10³⁰ solutions at least even if only flexibility of ligand is considered. Furthermore, database searching and rigid computing conditions are needed by this application. Problems mentioned above can be solved through using the techniques of grid.Development situation about drug molecular design is discussed first, then, techniques and applications on drug grid at present are discussed in this paper, such as data-transferring optimization, resources management optimization, task scheduling optimization, security and performance of docking software optimization. To overcome these problems, some methods related are developed. We briefly state our researches as follows:1. Based on the characters of ligand database's easy division and docking's independency, an optimized data-transferring method and model related are presented in this paper. It not only conquers the shortcomings of traditional methods but also make docking tasks distributed according to dynamic performance of every server.2. A dynamic and self-adaptive method is provided through using the virtues of market mechanism. The frame of resources management is provide firstly; then, a mathematic model related is presented and an algorithm using for resources distribution is designed; the shortcomings of market mechanism, such as too frequent fluctuation of price, are improved by means of changing prices after trading; using atom allocation of resource group, the cooperating allocation is improved, and some problems, such as deadlock of resource and inefficiently occupying resource, are avoided.3. A method RDHTS of grid task scheduling is presented in this paper to match the requirement of dynamism and heterogeneity of grid. Two obvious differences between RDHTS and traditional methods are: first, the matrix COST is designed to express communication fees between tasks and it's values are changed periodically according to changes of RTTs (Round Trip Time); second, the reliability of grid task scheduling is increased much through the coefficient of variation of reliability, which is obtained by analyzing the mean and variance of the number of the node's own sequential jobs, that's to say, it adapts the non-dedicated character of grid.4. An authority certificate CA is founded, which is responsible for distributing certificates. Then, distinguish the authenticity of certificates after CA distributed them according to mechanism of public key. Last but not at least, setting up mapping between grid name and local name of an entity, which is used to ensure the efficiency of local security mechanism.5. Methods and techniques mentioned above are applied in drug grid research. A dynamic distribution of docking tasks is set up through the quantity descriptions of node and domain performances. The optimized grid techniques, like, data-transferring optimizing method, task scheduling optimizing method and resources management optimizing method are realized in this paper. The security of platform is ensured through CA and mapping between grid userand local user. What's more, the reliability of docking is improved by analyzing the mean and variance of the number of the node's own sequential jobs. The results of example test have shown that the combination of drug molecular docking design and grid techniques is efficient and reasonable.This dissertation is financially supported by the National Natural Science Foundation (10 272030) and Subsidized by the Special Funds for Major State Basic Research Project of China (G1999032805, 2004cb518901).