KLAST: fast and sensitive software to compare large genomic databanks on cloud

Ivaylo Petrov 1 Sébastien Brillet 1 Erwan Drezen 1 Sylvain Quiniou 2 L Antin 2 Patrick Durand 2 Dominique Lavenier 1
1 GenScale - Scalable, Optimized and Parallel Algorithms for Genomics
Inria Rennes – Bretagne Atlantique , IRISA-D7 - GESTION DES DONNÉES ET DE LA CONNAISSANCE
Abstract : As the genomic data generated by high throughput sequencing machines continue to exponentially grow, the need for very efficient bioinformatics tools to extract relevant knowledge from this mass of data doesn't weaken. Comparing sequences is still a major task in this discovering process, but tends to be more and more time-consuming. KLAST is a sequence comparison software optimized to compare two nucleotides or proteins data sets, typically a set of query sequences and a reference bank. Performances of KLAST are obtained by a new indexing scheme, an optimized seed-extend methodology, and a multi-level parallelism implementation. To scale up to NGS data processing, a Hadoop version has been designed. Experiments demonstrate a good scalability and a large speed-up over BLAST, the reference software of the domain. In addition, computation can be optionally performed on compressed data without any loss in performances.
Complete list of metadatas

Cited literature [11 references]  Display  Hide  Download

https://hal.inria.fr/hal-01235339
Contributor : Dominique Lavenier <>
Submitted on : Monday, November 30, 2015 - 9:32:09 AM
Last modification on : Thursday, February 7, 2019 - 2:33:40 PM
Long-term archiving on : Saturday, April 29, 2017 - 12:03:18 AM

File

BioComp_BIC2743.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-01235339, version 1

Citation

Ivaylo Petrov, Sébastien Brillet, Erwan Drezen, Sylvain Quiniou, L Antin, et al.. KLAST: fast and sensitive software to compare large genomic databanks on cloud. World Congress in Computer Science, Computer Engineering, and Applied Computing, Jul 2015, Las Vegas,, United States. ⟨hal-01235339⟩

Share

Metrics

Record views

530

Files downloads

217