Introduction - User-Level Workflow Design: A Bioinformatics Perspective - page 24

service can not only be used by its direct successor, but is available to all sub-
sequently called services, which enables further service compositions. Control-
flow structures can only be modeled manually in the current framework, and
not be synthesized automatically. They are, however, fully recognized by the
data-flow analysis, which thus makes sure that all available data is taken into
account when the synthesis concretizes a loose branch at any place in the
workflow model.
As a result of these enhancements of the jABC framework, well-established
workflow management technologies are provided together with future-oriented,
semantic approaches to workflow design within one coherent framework. Its
broad range of applications includes complex, long-living applications like the
Online Conference Service (OCS) [238], as well as ad-hoc workflows [335] like
the exemplary bioinformatics applications, which can be dynamically adapted
to changing experimental requirements.
In fact, the implementation of the automatic workflow composition func-
tionality was essentially driven by the requirements identified when working
on bioinformatics workflows based on the jABC framework and related tech-
nologies. While the initially available, jABC-based comprehensive framework
for service-oriented, model-driven design, execution, verification and deploy-
ment of bioinformatics workflows has been coined Bio-jETI already by [201],
the concepts for its extension by functionality for semantics-based, (semi-) au-
tomatic workflow composition [174] are the result of the work that has been
carried out in the scope of the doctoral dissertation that constituted the
basis for this topic. Among bioinformatics workflow management systems,
Bio-jETI is unique in its application of different flavors of formal methods
for workflow design, which is facilitated by the formally defined structure of
the workflow models of the underlying jABC framework.
1.2.2 Applications and Evaluation
In order to explore the characteristics, capabilities and limits of the loose pro-
gramming approach to user-level workflow design, the framework described
above was applied to different real-life bioinformatics workflow scenarios.
Bringing it into application in a particular domain is in the first place a com-
prehensive domain modeling task: The jABC framework and the associated
technologies have been developed in a domain-independent fashion, provid-
ing general functionality for workflow modeling, such as a graphical editor
and mechanisms for persistence, execution, compilation, validation and de-
ployment. In order to use it for a specific application domain, it has to be
prepared for use in the target domain by selecting a suitable set of plugins and
providing adequate workflow building blocks (services) as well as semantic
meta-information about these services in terms of taxonomies or ontologies.
The Bio-jETI incarnation of the jABC framework that is used in the scope
of this topic is principally formed by its characteristic set of plugins. The do-
main models then tailor the available workflow building blocks and semantic