Taverna workflows can invoke general
SOAP/
WSDL or
REST Web services, and more specific SADI, BioMart,
BioMoby and SoapLab Web services. It can also invoke
R statistical services, local Java code, external tools on local and remote machines (via
ssh), do
XPath and other text manipulation, import a spreadsheet and include sub-workflows. Taverna Workbench includes the ability to monitor the running of a workflow and to examine the
provenance of the data produced, exposing details of the workflow run as a
W3C PROV-O
RDF provenance graph, within a structured
Research Object bundle
ZIP file that includes inputs, outputs, intermediate values and the executed workflow definition; together this format is called
TavernaProv. Taverna includes the ability to search for services described in
BioCatalogue to invoke from workflows. However, services do not need to be described within BioCatalogue to be included in workflows as they can be added from a
WSDL Web Service description or entered as a
REST URI pattern. Taverna also includes the capability to search for workflows on
myExperiment. The Taverna Workbench can download, modify and run workflows discovered on myExperiment, and also upload created workflows in order to share them with others using the social aspects of myExperiment. Taverna workflows do not need to be executed within the Taverna Workbench. Workflows can also be run by: • a
command line execution tool • remote execution server that allow Taverna workflows to be run on other machines, on computational grids, clouds, from Web pages and portals • online workflow designer and enactor
OnlineHPC Taverna allows pipelining and streaming of data. This means that services downstream in the workflow can start as soon as the first data item is received, without waiting for the whole data list to become available from upstream services and iterations. Taverna services execute in parallel when possible, as Taverna workflows are primarily data-driven rather than control-driven. == Open source community ==