Inria / Raweb 2004
Project-Team: MODBIO

Search in Activity Report, year 2004:
HELP

INDEX

Project-Team : modbio

Section: Software


KOALAB: KOupled Algorithmic and Learning Approach for Biological sequences

Participants: Damien Eveillard, Abdelhalim Larhlimi [correspondent], Sandrine Schermack-Peyrefitte [correspondent].

KOALAB is a software dedicated to biologists which has been developed to provide a user-friendly interface for the M-SVM (Multi-class Support Vector Machines) technology developed in the team. It is a tool based on web technology easy to use and necessitating only an Apache server to be installed. Its purpose is the search for regulatory motifs of biological interest within nucleic acids (ADN or ARN). The user can provide a collection of motifs on which the program will perform a learning process. There is no need to do any prior alignment of those motifs, a step necessary but limiting for other methods available. KOALAB also integrates the nucleic acid version of grappe, the motif finding algorithm developed by the ADAGE project-team. It hence offers the possibility to confront in the same graphical representation motif search results based on statistical learning with those obtained by the algorithmic methodology commonly used to date. A first study for two splicing regulatory protein targets on the HIV-1 virus genome (see Sect.  6.6) shows that this method is powerful compared to others available such as global consensus research or the software ESEfinder [30], which is based on Hidden Markov Models (HMM). KOALAB 1.0 can be downloaded from our website under the GNU GPL licence. It has been presented at JOBIM'2004 conference [19], and an installation on the bioinformatics server of LORIA is being under study.


previous
next