Phylogenetic Analysis Workflows - User-Level Workflow Design: A Bioinformatics Perspective - page 71

6. Saving the alignment to the local file system. Storing results of analyses
permanently rather than (only) showing them to the user at one point is
useful whenever they are relevant outside of the specific workflow.
7. Letting the user enter a keyword, which is then used for a DDBJ database
search (via the ARSA system) that results in a list of accession num-
bers for which the corresponding sequences are fetched from the DDBJ
database. As the input data for an alignment is usually not initially avail-
able at the local file system, this is an useful option for obtaining a set
of sequences with particular characteristics.
8. Extracting the phylogenetic tree that is part of a ClustalW result (using
a regular expression) and calling the phylip2jpg service of the DDBJ
that converts the textual tree representation into an image, followed by
retrieving and displaying the image.
These snippets can now be put together to various alignment workflows
simply by connecting them with the appropriate branches. For instance, con-
necting the snippets 1, 2 and 3 results in the basic alignment workflow from
above. Connecting the snippets 7, 2 and 8 forms a more complex workflow
(depicted in Figure 3.8), comprising database search by keyword, sequence re-
trieval, alignment computation, and visualization of the implied phylogenetic
tree.
Note that the examples shown here are really only some simple examples
of multiple sequence alignment workflows, suitable to give a first impression
of workflow modeling in Bio-jETI. They do not cover all the modeling ca-
pabilities of the system. The features shown in the different versions of the
example can be combined, and the workflows can be extended with whatever
functionality is available and encapsulated in a SIB. For instance, the DDBJ's
BLAST web service can be used to retrieve a set of sequences based on sim-
ilarity to a query sequence rather than a query keyword as in the example
above. A corresponding workflow is among the DDBJ's workflow examples
[4], which list a number of workflows that make use of the DDBJ services.
And naturally also other institutes' data retrieval and sequence analysis ser-
vices can be applied, for instance those provided by the EBI [250, 167, 118]
or BiBiServ [127].
3.3 Constraint-Driven Design of Phylogenetic
Analysis Workflows
This section describes and discusses a scenario for the loose programming
of phylogenetic analysis workflows based on a large and complex domain
model that comprises similar, but much more services than those described in
the previous section. This comprehensive example is particularly suitable to
demonstrate the advantages of (semi-) automatic, semantics-driven workflow
composition.