Related Work - User-Level Workflow Design: A Bioinformatics Perspective - page 179

further, simple dependencies can be expressed by defining additional control
links , or by defining iteration strategies for single components.
Taverna workflows can directly be executed once all involved services are
connected. The workflow execution is managed in a separate window, in the
so-called Results Perspective, which takes care of workflow execution, doc-
uments the progress and status of its execution in detail, and presents the
(final and intermediate) results. Upon execution, each of the specified work-
flow inputs has to be provided with data, the outputs are then listed in the
results perspective, from where they can be processed further.
Example 1 in Taverna
Coming to the realization of the workflow for analysis of multiple sequences in
Taverna, the first step is to search the list of available processors for the appro-
priate EMBOSS tools. As Taverna's discovery mechanism browses the services
provided by EBI's SoapLab [14, 15] automatically, they are directly available
and can be added to the model. Furthermore a a workflow input component
is needed that represents the multiple sequences that the workflow deals with,
and three output representatives for the consensus sequence, the HMM and the
emitted sequences.
Fig. 8.8 Multiple sequences analysis workflow in Taverna
The data links are defined as can be seen in Figure 8.8: The initial se-
quences are sent to emma , the resulting alignment is the input for both cons
and ehmmbuild , and the constructed HMM is sent to ehmmcalibrate and
ehmmemit .For ehmmbuild a name for the constructed HMM must be spec-
ified, which is done via the String constant processor hmm name . The finally
resulting consensus sequences and HMM-emitted sequences are directed to
the workflow output components consensus and sequences , respectively.
Example 2 in Taverna
Fig. 8.9 Phylogenetic tree construction workflow in Taverna