Web services and workflow management for biological resources

Abstract

Background: The completion of the Human Genome Project has resulted in large quantities of biological data which are proving difficult to manage and integrate effectively. There is a need for a system that is able to automate accesses to remote sites and to "understand" the information that it is managing in order to link data properly. Workflow management systems combined with Web Services are promising Information and Communication Technologies (ICT) tools. Some have already been proposed and are being increasingly applied to the biomedical domain, especially as many biology-related Web Services are now becoming available. Information on biological resources and on genomic sequences mutations are two examples of very specialized datasets that are useful for specific research domains.

Results: The architecture of a system that is able to access and execute predefined workflows is presented in this paper. Web Services allowing access to the IARC TP53 Mutation Database and CABRI catalogues of biological resources have been implemented and are available on-line. Example workflows which retrieve data from these Web Services have also been created and are available on-line.

Conclusion: We present a general architecture and some building blocks for the implementation of a system that is able to remotely execute workflows of biomedical interest and show how this approach can effectively produce useful outputs. The further development and implementation of Web Services allowing access to an exhaustive set of biomedical databases and the creation of effective and useful workflows will improve the automation of in-silico analysis.

Figure 1

General architecture of the system. The general architecture of the system includes three blocks: workflows creation and annotation (WCA), user interface (UI) and workflows execution (WE). The "Taverna" and "FreeFluo" boxes refer to open source software developed externally from the project. The boxes named "Repository Manager", "Workflow Executor" and "Interface" refer to purpose applications developed within the project. All other boxes refer to databases, either local or remote.

Figure 2

ACD definitions for two Soaplab based Web Services searching CABRI cell lines catalogues. This figure shows the definitions used for the set-up of the Web Services that return unique identifiers of cell lines in CABRI catalogues after a search either by name or by property (free text search). These definitions are written in the ACD language and must be converted into XML before they can be used. Differences between the two definitions consist in the comment line where the actual parameters for the remote call to the CABRI site are defined

Figure 3

Diagram of the GetBacteriaByName workflow. This diagram lists one of the workflows that have been created and are available on-line for demonstration purposes. Two web services are used (getBacteriaIdsByname and getBacteriaById), together with some local string elaborations. This output has been generated by the Taverna Workbench.