BLAST
Partnership
Web site:
Context
BLAST is a popular bioinformatics application used to find similarities between nucleotids or amino-acids sequences. The aim of such a research is to find some clues about the function of a given unknown gene or protein. The new sequences are compared to sequences stored into large biological databases which can also contain annotations about their functions and references to publications about them.
Use case
The SysFera-DS BLAST application was designed to manage thousands of BLAST requests over large biologocal databases. We tested it, submitting 40000 requests over 5 databases of different sizes (from 1 to 5 GB). Each request is submitted asynchronously to the platform, SysFera-DS selecting the computations node using a scheduling algorithm. We tested four different scheduling policies: a simple greedy algorithm, MCT, SRA and dynamic-SRA.
Contribution
- Experiments on the Grid'5000 platform involving up to 1000 computing nodes obtaining the results for the 40000 in few minutes.
- Up to 30% gains on completion time using the dynamic-SRA algorithm.
Technical Form
| Type | Valeur |
|---|---|
| Code | Parallel on SMP systems |
| CPU itensive | Medium |
| I/O intensive | High |
| Memory intensive | High |
| Plug-in scheduler | Yes |
| SysFera-DS Client/Server | Available |
|
Deployment |
Grid'5000 |
References
-
V. Breton, E. Caron, F. Desprez, G. Le Mahec. High Performance Data Management and Scheduling for BLAST Applications over the Grid - From Parallel Algorithms to Middleware Issues. Handbook of Research on Computational Grid Technologies for Life Sciences, Biomedicine and Healthcare. Mario Cannataro, University Magna Graecia of Catanzaro, Italy (editor) - Publisher : Information Science Reference. 2009.
- E. Caron, F. Desprez et G. Le Mahec. Parallelization and Distribution Strategies of Large Bioinformatics Requests over the Grid. Proceedings of ICA3PP’2008, LNCS. International Conference on Algorithms and Architectures for Parallel Processing 2008 : Cyprus - June 9-11 2008.
