Distributed Problem Solving Environment Dedicated to DNA Sequence Annotation

Genomic sequence analysis is a task using techniques coming from different fields in order to extract biologically relevant objects (genes, regulatory signals...) from rough DNA sequences. Analysis methods, coming from domains like statistics, sequence alignment or pattern matching, have been developed and regrouped into program libraries. These libraries aim at helping biologists to manipulate data. However their use revealed itself being too difficult as it requires the user to have background knowledge to be handled efficiently. In order to tackle this problem, the ImaGene system has been proposed. ImaGene is a system built upon a generic task model allowing to model methods of DNA sequence analysis and to execute them thanks to shell scripts, binaries and specific libraries already developed. This system makes it possible to manipulate these methods and present a synthesis of the obtained results in a cartographic interface allowing the biologist to evaluate the biological pertinence of the results and to annotate DNA sequences. In order to profit from analysis methods and specialized libraries and to simplify their accesses, we have provided ImaGene with the possibility of using distributed methods and binaries. The solution chosen consists in transferring task code from server sites to client sites; but, unlike the solution adopted in Java, the data is computed at the server site if necessary. This distribution of the system has highlighted some problems like managing different versions of methods and dealing with tasks that have been recently decomposed into new sub-tasks.