Taverna was the largest project I worked on at the EBI. Aimed at (non-computer) scientists it allows the user to construct branching and merging 'pipelines' for data analysis. The components in these pipelines can be local workers such as scripts, but are predominantly remote resource such as SOAP web services, data warehouses and a variety of community standards for service provision such as BioMoby. One of the key challenges therefore was to create a client which would work with any service it found, this implied that the client must be dynamic, something most client APIs were and are poorly equipped to handle.
In addition to the problem of actually making the pipelining system work with these external resources we had to make it usable for non-expert users. This involved the construction of a sophisticated rich client interface including a visual representation of the data processing pipeline. A screenshot of version 1.7 of the workflow workbench is shown below:
Taverna is still in active development by the team at the University of Manchester, and can be found at www.taverna.org.uk