BioMoby extensions to the Taverna workflow management and enactment software

Abstract

Background: As biology becomes an increasingly computational science, it is critical that we develop software tools that support not only bioinformaticians, but also bench biologists in their exploration of the vast and complex data-sets that continue to build from international genomic, proteomic, and systems-biology projects. The BioMoby interoperability system was created with the goal of facilitating the movement of data from one Web-based resource to another to fulfill the requirements of non-expert bioinformaticians. In parallel with the development of BioMoby, the European myGrid project was designing Taverna, a bioinformatics workflow design and enactment tool. Here we describe the marriage of these two projects in the form of a Taverna plug-in that provides access to many of BioMoby's features through the Taverna interface.

Results: The exposed BioMoby functionality aids in the design of "sensible" BioMoby workflows, aids in pipelining BioMoby and non-BioMoby-based resources, and ensures that end-users need only a minimal understanding of both BioMoby, and the Taverna interface itself. Users are guided through the construction of syntactically and semantically correct workflows through plug-in calls to the Moby Central registry. Moby Central provides a menu of only those BioMoby services capable of operating on the data-type(s) that exist at any given position in the workflow. Moreover, the plug-in automatically and correctly connects a selected service into the workflow such that users are not required to understand the nature of the inputs or outputs for any service, leaving them to focus on the biological meaning of the workflow they are constructing, rather than the technical details of how the services will interoperate.

Conclusion: With the availability of the BioMoby plug-in to Taverna, we believe that BioMoby-based Web Services are now significantly more useful and accessible to bench scientists than are more traditional Web Services.

Figure 1

A BioMoby workflow. The function of this workflow is to retrieve the list of loci from Antirrhinum (Snapdragon) and Arabidopsis which share both a keyword in their phenotypic description as well as sequence homology of the affected locus, thus more likely representing true homologues. Yellow boxes represent BioMoby Objects; orange boxes represent BioMoby services; white boxes represent BioMoby Object parsers; Purple boxes represent local Java Bean widgets; Blue boxes represent inputs and outputs.

Figure 2

Taverna's Available Services Menu. A BioMoby service "KEGG2Frequencies" has been selected and is about to be added to the workflow model from the pop-up menu. This is how services are traditionally added to a workflow in Taverna.

Figure 3

Connecting Services using the Advanced Model Explorer. The output port from the service getAminoAcidSequence is about to be connected to the input port of the GenericSequence2FASTA service. This is accomplished by selecting the output port of the preceding service, right-clicking, and selecting the desired output port from the cascade of pop-up menus.

Figure 4

The BioMoby Object Ontology as widgets within the Available Services Menu. The BasicGFF3SequenceFeature widget is about to be added to the workflow.

Figure 5

The Moby Service Details option in the Advanced Model Explorer. Unlike non-BioMoby services, the ability to retrieve more advanced details about BioMoby services is made available by the BioMoby plugin as a right-click menu option.

Figure 6

Adding a BioMoby output parser. The output data from any given service can be fed directly into an appropriate XML parser simply by selecting the output data-type from the Moby Service Details window and choosing the "Add a parser" menu option.

Figure 7

The Configure Moby Service menu option. The BioMoby plugin recognizes when a BioMoby service has additional configuration parameters, and adds an additional menu option to the Advanced Model Explorer's right-click menu.

Figure 8

The Configure Moby Service window. All configurable parameters are presented for manual manipulation. In this case, a parameter called "genome" is presented, which is of an "ENUM" data-type, where the different acceptable genomes are listed. The Yeast genome has been selected.

Figure 9

The Moby Service Discovery menu. Unlike any other service in Taverna, the output of a given service can be used to automatically identify all services that are capable of consuming that output, and thus representing the next possible steps in the workflow. There are two ways to achieve the search – with traversal of the Object Ontology (semantic search) or without (brief search). Here, the brief search option has been selected.

Figure 10

The output from a service discovery. The Object Dertails window displays a list of services capable of consuming the output object from the preceding service. These are organized by service provider (red icons) and all services from a given provider that are capable of consuming that output are presented for selection (purple icons). A mouse-over pop-up help window is visible describing the functionality of the PhytoprotGetByAminoAcidSequence service provided by urgi.versailles.inra.fr.