Related Experiment Videos
DPRml: distributed phylogeny reconstruction by maximum likelihood.
T M Keane1, T J Naughton, S A A Travers
1Department of Computer Science, National University of Ireland Maynooth, Ireland.
Bioinformatics (Oxford, England)
|October 30, 2004
Summary
We developed a distributed phylogeny reconstruction program that leverages idle computing resources to build large phylogenetic trees efficiently. This approach offers a cost-effective solution for complex phylogenetic analyses, achieving near-linear speedup.
Area of Science:
- Computational Biology
- Bioinformatics
- Phylogenetics
Background:
- Increasing demand for large and accurate phylogenetic trees using statistical methods.
- Limitations of single-processor systems for constructing trees with numerous taxa.
- Concerns regarding the efficiency and accessibility of existing parallel phylogenetic programs.
Purpose of the Study:
- To assess the suitability of phylogenetic analysis for large-scale, heterogeneous distributed computing.
- To develop a novel distributed program for maximum likelihood-based phylogenetic tree construction.
Main Methods:
- Developed a distributed, cross-platform program: distributed phylogeny reconstruction by maximum likelihood.
- Utilized a proven maximum likelihood algorithm and a popular phylogenetic analysis library.
- Leveraged idle clock cycles from networked machines for computation.
Main Results:
- Achieved near-linear speedup in phylogenetic tree building using distributed computing.
- Demonstrated the feasibility of creating a 'free' supercomputer for maximum likelihood analysis in academic/corporate settings.
- The program supports an extensive range of DNA substitution models.
Conclusions:
- Distributed computing is highly suitable for large-scale phylogenetic analysis.
- The developed program offers an efficient and accessible solution for complex tree reconstruction.
- Idle computing resources can be effectively harnessed for significant computational gains in phylogenetics.