A data decomposition middleware tool with a generic built-in work-flow

Authors:
Ahmad Salah;Kenli Li
Affiliations:
Hunan University, Changsha, Hunan, China and Zagazig University, Zagazig, Egypt and Hunan University, Changsha, Hunan, China;Hunan University, Changsha, Hunan, China
Venue:
Proceedings of the 20th European MPI Users' Group Meeting
Year:
2013

Citing 4
Cited 0

Advanced resource connector middleware for lightweight computational Grids

Future Generation Computer Systems - Special section: Information engineering and enterprise architecture in distributed computing environments
ProBiS algorithm for detection of structurally similar protein binding sites by local structural alignment

Bioinformatics
A Grid-Enabled Gateway for Biomedical Data Analysis

Journal of Grid Computing
Distributed Application Runtime Environment (DARE): A Standards-based Middleware Framework for Science-Gateways

Journal of Grid Computing

Quantified Score

Hi-index	0.00

Visualization

Abstract

The steady increase of the biological data encourages computer scientists to develop data-analytical methods in order to study the biological systems. Most of these methods are firstly presented in the sequential version and then it is converted to a parallel version due to the constant expanding of data size. In this paper, we present a tool to eliminate the time gap between the two versions, sequential and parallel. The proposed tool acts as a middleware with a generic built in work-flow. The middleware manages a set of instances of the same sequential method; it manages to assign the data to each instance, to merge the results of all instances, and to form the final results. The proposed tool is applicable for problems with data decomposition nature (e.g., protein structure comparison, sequence comparison and CpG Islands search). It targets two levels of parallelism, which are the homogenous multi-core processor architectures and networks of heterogeneous computational nodes. The experimental results, of different computational biology problems, show a speedup close to the optimal with identical accuracy.