You are in the accessibility menu

Please use this identifier to cite or link to this item: http://acervodigital.unesp.br/handle/11449/122821
Title: 
Using Threads to Overcome Synchronization Delays in Parallel Multiple Progressive Alignment Algorithms
Author(s): 
Institution: 
  • Universidade Estadual Paulista (UNESP)
  • Universidade de São Paulo (USP)
  • Zhejiang University
ISSN: 
1948-9862
Sponsorship: 
Sao Paulo State Research Foundation-FAPESP
Sponsorship Process Number: 
FAPESP: 06/59592-0
Abstract: 
Problem statement: The parallelization of multiple progressive alignment algorithms is a difficult task. All known methods have strong bottlenecks resulting from synchronization delays. This is even more constraining in distributed memory systems, where message passing also delays the interprocess communication. Despite these drawbacks, parallel computing is becoming increasingly necessary to perform multiple sequence alignment. Approach: In this study, it is introduced a solution for parallelizing multiple progressive alignments in distributed memory systems that overcomes such delays. Results: The proposed approach uses threads to separate actual alignment from synchronization and communication. It also uses a different approach to schedule independent tasks. Conclusion/Recommendations: The approach was intensively tested, producing a performance remarkably better than a largely used algorithm. It is suggested that it can be applied to improve the performance of some multiple alignment tools, as CLUSTALW and MUSCLE.
Issue Date: 
2012
Citation: 
American Journal of Bioinformatics, v. 1, n. 1, p. 50-63, 2012.
Time Duration: 
50-63
Keywords: 
  • Sequence alignment algorithms
  • Hidden Markov Models (HMM)
  • performance remarkably
  • progressive approach
  • computational complexity
  • represent genetic
Source: 
http://thescipub.com/abstract/10.3844/ajbsp.2012.50.63
URI: 
Access Rights: 
Acesso restrito
Type: 
outro
Source:
http://repositorio.unesp.br/handle/11449/122821
Appears in Collections:Artigos, TCCs, Teses e Dissertações da Unesp

There are no files associated with this item.
 

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.