Desiderata for TavernaWorkbench

Check out the User requirements and Technical requirements

New Events

The following events would help improve provenance capture:

1. NestedWorkflowCreationEvent (with a reference to the creating process)
2. ProcessCreationEvent (useful for performance evaluation)

WorkFlow Repository

The provenance plugin provides a way to resolve several kinds of LSIDs assigned by TavernaWorkbench, but not LSIDs of workflows. Eg:

• urn:lsid:www.mygrid.org.uk:operation:16OM6H01JJ0

A repository is needed in order to resolve such LSIDs.

The repository could also be used to keep track of reused workflows (embedded workflows)

Datastore

• Give TavernaWorkbench's datastore the option to store references to dataElements rather than values.
• Ability to store only user-tagged data
  • tag data as interesting
    • tag eg only third element in a list as interesting

Security

• Mark sensitive parts of the workflow users want to restrict access to.
  • issue private/restricted LSIDs

LSIDs

• One data/one LSID instead of assigning new LSIDs to the same data. Pass LSID references to Taverna as inputs.
• Choose which (meta)data we want to publish as LSIDs. Refer to the rest using generic URNs.

Configuration

• Browse and edit mygrid.properties from Taverna

Adding new services to the registry

Currently new services are added to the registry in an ad hoc manner, this process needs to be formalised, so that new services can be added or updated as appropriate. For example, as of January 2006 some of the following services are not available in the default list of services, but probably should be.

• InterProScan
• DBFetch
• ClustalW
• TCoffee
• Fasta
• WUBlast2
• MPSearch
• ScanPS
• Pfam Web Services
• PRIDE web services

BeanShell^? scripting host

The BeanShell^? environment in Taverna needs better debugging features if it is to be at all useful to end-users. E.g. give line numbers, if there is an error in the script rather than outputting all on one line.

Multiple workflow editing

The ability to load multiple workflows in Taverna at the same time, and to move (select, copy, remove) large chunks of workflow across these workflows.

General Requirements

• Complex types
  • examples:
    • InterProScan
    • ScanPS
    • ClustalW
    • TCoffee
    • Fasta
    • WUBlast2
    • MPSearch
    • Emboss
    • Muscle
• Performance/scalability
  • stream-based model as already announced
    • long execution times
    • large scale of data processed in parallel
  • we should collect break points
• Developer-friendly API
  • components/development framework
  • developers guide
• Complex conditional flow
  • is there any user requirement for this?

GUI (users requirements)

• Available Services window
  • put property that prevents expansion
  • possibility to save current state
  • refresh selected providers
• Connect Workflow diagram to Advanced model explorer
• save current layout
• GuiRequests

Documentation framework for processors

A managed approach to dealing with multiple sources of information describing processors. These sources include:

• underlying service type, e.g. whatever's captured in a WSDL definition
• semantic information: service annotation with Feta
• workflow information: as part of the workflow building, services get renamed/described in natural language. One can see this as providing contextual info to a service, i.e. it has been used in this particular workflow fragment and in this context was named such and such.

See also WorkflowDocumentationTemplate

See Also

• Taverna2 thoughts