Conclusion - User-Level Workflow Design: A Bioinformatics Perspective - page 195

involved entities (cf., e.g.,[68]) also the development of specific domain
ontologies has already started (cf., e.g., [333, 194]).
The application scenarios described in this topic are local in the sense the
they deal with particular data analysis workflows, such as the design of PCR
primers (cf. Chapter 5) or the analysis of data from metabolic flux experi-
ments (cf. Chapter 4). They are, however, usually parts of larger research pro-
cesses: Complex genomics and expression profiling experiments, for instance,
require the amplification of DNA fragments and thus suitable primers for
the PCR at different points. Or, as another example, the development of dia-
betes treatments needs detailed knowledge about cell metabolism (which can
be provided by the flux experiments), but also involves different other fields
of work, such as the actual design of pharmaceuticals, the clinical trials and
the drug licensing procedures.
This suggests to extend the application scope of the developed method-
ologies from the currently addressed local data analysis workflows to global
research management processes. These larger, typically hierarchical processes
can as well be modeled with the framework in order to optimize the research
work from a more global perspective. This may e.g. comprise the automatic
comparison with previous analysis results, the generation of reports, as well
as an adequate archiving of the results for later reuse.
On the one hand, this is facilitated by the flexibility of the framework with
regard to the concrete applications. In fact, the characteristics of the individ-
ual applications are essentially captured by the respective domain models.
Hence, the methods can be expected to apply also to the management of the
higher-level research management processes, which do in fact have a business-
process flavor rather than the typical characteristics of scientific application
domains. On the other hand, the hierarchical control-flow structure of the
workflow and process models is central with regard to achieving this goal.
Hierarchical models are required to structure complex processes adequately
and represent them at different levels of detail/granularity, according to the
focus of the respective user. The control-flow structure of the models allows
for a more decisive representation of process flows than data-flow modeling
could achieve, and by being less focused on the data, it is also more suitable
for modeling the higher-level processes of complex research projects.
Finally, it is especially the application of a coherent modeling formalism
throughout the framework which will facilitate the adoption of the methodol-
ogy by the different users. In combination with the specific domain models that
allow the users to use the terms oftheir respective technical languages,they will
enable the users to effectively work at the various areas of expertise and levels
of granularity. Thus, true user-level handling of all local workflows and global
processes involved in complex research projects will become possible.