Font Size: a A A

Research On Grid Workflow And Its Application In Drug Discovery

Posted on:2008-11-30Degree:MasterType:Thesis
Country:ChinaCandidate:X A DingFull Text:PDF
GTID:2178360242976770Subject:Computer applications
Abstract/Summary:PDF Full Text Request
Drug Discovery is a very important task in human health area. Nowadays, more and more countries and companies are paying attention to the Computer Aided Drug Discovery (CADD). The process of CADD refers to many software packages and databases, which are closely related to each other. Therefore, in order to create different drug discovery workflows, many kinds of models,theories and databases should be packed into large-scaled,port-to-port applications, and should also be integrated according to their logic relations hip too.However, it isn't so easy to integrate these applications. On one hand, the upgrade and improvement of applications will be hard because of integrating the codes; on the other hand, integrating the codes is so hard for the non-professional end users. Luckily, current grid workflow system can satisfy this requirement, which can provide the foundational support for end users to self-define and run the drug discovery flows.At first, the architectures of Workflow System for Drug Discovery Grid (WSfDDG) and Grid Workflow Management System (GWfMS) are introduced in this article.Secondly, in order to improve scalability and stability of the WSfDDG, an automatic configuration algorithm for distributed workflow engine based on fuzzy control is brought out in this paper, because current simple and static configuration mechanism doesn't work well when the load of GWfMS changes greatly with the time.Thirdly, the scheduling policies of grid workflow running are discussed for the commercial grid and the non-commercial one. And what's more, a new grid task-service mapping algorithm, also named"A Fuzzy Control based Dispatch Algorithm for Grid Tasks", is brought out to decrease the average running time of the most grid tasks.And then, a task-partition algorithm for WSfDDG is brought out to improve the performance of Drug Docking. The algorithm is also compared with the one of mean task-partition.Finally, a developed prototype WSfDDG is introduced. Jini technology is applied to develop this prototype system. The experiment shows the flexibility, extensibility and reliability of the GWfMS have been improved. And what's more, we implemented a grid middleware prototype for grid workflow scheduling, and the test environment for our task-partition algorithm is also proposed.The experiment shows that this WSfDDG possesses the strong abilities of service automatic configuration,optimized scheduling for grid services,node-balanced task partition. The automatic configuration mechanism improves the ability of adjustment and configuration to adapt to the changing environment, and the new grid task-service mapping algorithm can reduce the running time of the grid workflow. Because of its usability, reliability and extendability, the WSfDDG will be widely used.
Keywords/Search Tags:Drug Discovery, Grid Workflow, Service Configuration, Fuzzy Control, Flow Scheduling, Task Partition, Jini
PDF Full Text Request
Related items