27 Aug 2006 For Taverna information go to TavernaWorkbench, for general workflow try WorkflowLinks

Contents

24 Jan 2005 Taverna Release 1.0 available, with user manual, see http://taverna.sf.net. This page has not been substantially revised since this first release - lots has happened since see TavernaWorkbench

Introduction

myGrid workflows provide a representation of in-silico scientific experiments (or more accurately the materials and methods parts of in-silico experiments). The workflows can be shared, ammended, indexed etc. allowing a community to share not just their results but also their processes.

The Taverna workbench is used to create, edit and run myGrid scientific workflows. For details and documentation see the Taverna web site. The Taverna workbench incorporates a version of the Freefluo enactment engine. Taverna has a plug-in facility that allows interaction with other myGrid components, in particular the information repository (MIR) for storing the results and context of in silico experiments and Feta for semantic search of service descriptions.

The myGrid workflow enactment engine is a workflow orchestration tool for web services. This is now open source and available on sourceforge , see http://freefluo.sourceforge.net/ ( Previous URL for the workflow enactment engine http://www.it-innovation.soton.ac.uk/mygrid/workflow/ )

While Taverna (and Freefluo) are not fundamentally linked to BioinformaticsWebServices the development has been driven by Bioinformatics users and their requirements for composing bioinformatics resources both locally and on the web.

Useful Links

• WorkflowLinks - general links of scientific workflows, semantic web services,
• Taverna (includes Scufl language) http://taverna.sourceforge.net
  • TavernaTips contains additional documentation including links to papers and presentations
• Workflow Enactment Engine, Freefluo http://freefluo.sourceforge.net (formerly at http://www.it-innovation.soton.ac.uk/mygrid/workflow )
• BioinformaticsWebServices (Tom's list http://www.ebi.ac.uk/~tmo/mygrid/webservices/ )
• BioinfWorkflows example Scufl workflows by StefanEgglestone
• Example XScufl workflows at WorkFlow#XScufl_workflow_definitions below
• NeSC workflow workshop 3-5 Dec 2003
• Chris Wroe's Workflow presentation for biologists and bioinformaticians, July 2004 (from PresentationStore)
• WorkflowAtGGF
• SoapLab http://industry.ebi.ac.uk/soaplab/ creating web services from command line applications
• Talisman http://talisman.sourceforge.net/
• FreefluoXScufl configuring the freefluo enactment engine for XScufl

Differences between eScience workflows and eBusiness

The document at WorkFlowDifferences is an attempt to summarize the differences between eScience workflows and their eBusiness counterparts. Mainly talks about workflow topology and lifecycle, it could do with more detail kinds of data manipulated as well. [Tom]

WorkflowLinks provides links to example scientific workflow systems and related discussions

Taverna Scufl Workbench

The images below are from an early version, and there have been significant improvements in response to users' feedback. For the latest version see the User Manual or the web site http://taverna.sf.net.

• Editor for Scufl, based on Java but uses some native extensions such as dot from AT&T
• Downloadable from cvs on sourceforge alongside the rest of the taverna project.

Semantic Markup in Taverna

• An example workbench session showing the user browsing an ontology and assigning the correct (we hope!) input type to the workflow input, note that this then appears in the XML view.

• There's also a string constant processor in the workflow with the default value 'MEDLINE' being fed into the data base cross referencing tool

XScufl workflow definitions

A reasonable collection of XScufl workflows are on the Taverna cvs and included in the Taverna workbench release.

• http://taverna.sourceforge.net/webservices/workflows/
• http://cvs.sourceforge.net/viewcvs.py/taverna/taverna1.0/workbench/examples/

XScufl workflow definitions for the Graves disease scenario can be found at: http://cvs.mygrid.org.uk/scufl

http://homepages.cs.ncl.ac.uk/peter.li/home.formal/scuflbeta6/ for workflows for Taverna beta6 Scufl, including semantic markup (October 2003)

These have been developed in the context of the Graves Disease study. Earlier examples may still be available at http://homepages.cs.ncl.ac.uk/peter.li/home.formal/workflow_definitions/scufl/

WilliamsTestWorkflows - longer running workflows from the work on Williams Syndrome, uses Blast and processing of Blast output

BlastTestWorkflows - specifically Blastn

AmbitTestWorkflows has some test workflows for the Ambit TextExtraction service

TempWSDL

Scufl

• Project page at http://www.sourceforge.net/projects/taverna
• Includes visualisation tools using dot from AT&T
• Should allow all workflows currently possible with WSFL plus some extra ones we couldn't do before.
• Consumes arbitrary web services (green), Talisman tscript services (purple) and Soaplab analysis tools (pale yellow)
• Allows for specification of explicit coordination constraints
• User documentation at http://sourceforge.net/docman/display_doc.php?docid=16697&group_id=74874
• See the 'hybridworkflow' directory inside the taverna1.0 module for more details, a direct link to the wincvs browser for this directory is http://cvs.sourceforge.net/cgi-bin/viewcvs.cgi/taverna/taverna1.0/hybridworkflow/ , this includes the talisman definition and script files which are accessed at runtime from www.cyclonic.org, as I needed somewhere to put them that the enactor could see via http. It also includes the workflow file itself, and the generated dot files and images created by the DotView.

-- TomOinn - 02 May 2003

Scufl and WSFL

The myGrid project initially developed workflows in IBM's WSFL (Web Services Flow Language), but in May-June 2003 changed to using Scufl. Scufl is a simple workflow language, with a focus on bioinformatics, that has been developed on the open source Taverna project ( http://taverna.sourceforge.net ) involving myGrid, in collaboration with the European Bioinformatics Institute and the Human Genome Mapping Project

In summary, with Scufl bioinformaticians are able to focus on the problem, the in-silico experiments, while with WSFL they had to grapple with the solution, the coordination of web services. For more detail see the Taverna project website, http://taverna.sourceforge.net , or the discussion documents on the Taverna sourceforge pages, http://sourceforge.net/docman/index.php?group_id=74874

Note the XML version of Scufl, XScufl, is now used extensively.

WSFL Workflows (pre-June 2003, now using Scufl)

WSFL Workflow Examples

• ExampleBioFetchWorkflow - get entry from SWISS-PROT
• ExampleEmbossWorkflow
• ExampleXemblWorkflow

WSFL Workflow Tests

• EmbossTests - including blast tests
• ConnectionTests
• MirTests

WSFL Workflow Resources

• WorkflowResources - XSL stylesheets for workflows and provenance records
• WorkflowTips - examples of problems

-- MarkGreenwood - 26 Mar 2003

Workflow Enactment Interactions

Available below.The archive includes the interaction template used for the definitions and an index page 'index.html'. You can extract the archive and navigate the interaction definitions using hyperlinks. It is proposed that interaction definitions are stored in CVS so that versioning can be provided and references to actual sample code is possible.

-- DarrenMarvin - 21 Feb 2003

• enactor_interactions.zip: Interactions associated with Workflow Enactment
• enactor_interaction_2.zip: Interactions associated with Workflow Enactment

• ECG2005WorkflowMScottMarshall.ppt: Scott Marshall's wish list for Taverna