Semi-automatic composition of bioinformatics Web services based on task and domain ontologies

Nowadays, there is a strong demand from biologists for tools to help them analyze biological experiments data. Despite the diversity of bioinformatics Web services used to process data and the ongoing development of platforms, the realization of a biological task remains a difficult problem. Overall, Web services composition is done by using workflows and currently done manually by users. It is desirable to improve the composition of Web services. However, automating a workflow is difficult. It requires to describe what Web services do as well as the nature of their input and output parameters. The aim of this survey is first to propose a semi-automatic composition of Web services, and second to obtain a better definition of bioinformatics Web services by modeling task and domain ontologies. As a proof of concept, we demonstrate a showcase workflow involving the combination of two Web services in order to realize a similarity search from an UniProt identifier. The first service getFASTA_UNIPROTEntry1 provides a protein sequence from an UniProt accession number. The second service searchSimple2 realizes a similarity search from the protein sequence obtained by the first service. This workflow can be transposed to Grids [1]. First, we created a task and a domain ontology. We enriched OWL-S [2] that allows the description of the task accomplished by a service, his functional mode and his grounding. We specified a task carried out by a service, by specializing the classes of OWL-S hierarchy. For example