Information Technology Reference
In-Depth Information
Example 3 in Bio-jETI
Fig. 8.6 DDBJ-Uniprot workflow in Bio-jETI
The third example is based on the DDBJ services, which have already been
used in the examples of Section 3.2. At the beginning of this workflow (top
left), an empty table is initialized, where during the workflow run the results
are collected. Then, the ARSA search service is called and the database IDs
of the matches are extracted from the result. The iteration over these IDs
constitutes the first loop. Within this loop, the entry corresponding to the
current ID is fetched and “blasted” against a database, again resulting in a
list of entries (represented by their IDs). Iterating over these IDs yields a
loop within a loop. Here, the database entries corresponding to the BLAST
hits are successively retrieved, and their IDs and descriptions are added to
the result. The resulting table finally consists of the IDs returned by ARSA
(first column), the IDs found by the BLAST search (second column) and
the corresponding ID and description fields of the actual database entries
(remaining columns).
8.2.3 Data-Flow Realization
For the realization of the benchmark workflows in a data-flow based environ-
ment, Taverna version 2.2.0 (released in July 2010) was used. Very similar
Bio-jETI, Taverna supports graphical workflow modeling on a central canvas,
where services are orchestrated into workflows. Figure 8.7 gives an impression
of Taverna's graphical user interface for the workflow construction (the so-
called Design Perspective): The design window consists of three basic parts: a
list of the available local and remote services (top left), a table managing the
services and links that constitute the workflow model (bottom left), and a
graphical representation of the workflow on the right. In contrast to Bio-jETI,
the branches between the (input and output ports of the) single services de-
fine a data-flow model of the process. In order to influence the flow of control
 
Search WWH ::




Custom Search