Introduction - User-Level Workflow Design: A Bioinformatics Perspective - page 28

the combinatorial blow-up of the state space [324], is an inherent issue of the
synthesis method. Although state explosion can not be entirely eliminated,
it can be effectively postponed by appropriate domain design and constraint
application. Thus, proper semantic domain modeling is obviously crucial. At
the same time, adequate domain modeling is clearly dicult, since it has to
take into account a plethora of aspects and furthermore depends massively
on the characteristics of the concrete application domain.
In this light, the loose programming pragmatics formulated in this topic
summarize the experiences gained by working on the application scenarios
and their evaluation and provide useful general guidelines for adequate do-
main modeling and successful synthesis application. In short, they state that:

1. Services and data types of adequate granularity are required as basis for
any domain model and workflow application.
2. Semantic domain modeling has to focus on the definition of a precise
domain vocabulary and the definition of simple semantic service interface
descriptions .
3. For the actual workflow design, it is advisable to increase the search depth
gradually and to specify the constraints incrementally .

Note that as in all software engineering processes, finally a good amount
of experience is required to obtain adequate solutions, and that also the
most carefully designed domain model can not be expected to suit all possi-
ble application scenarios equally well. A distinguishing feature of the loose
programming framework in this regard is that it explicitly encourages the
workflow designer to bring in his specific domain knowledge and adapt the
domain model according to his particular needs.

1.3 Outline

Concretely, the remainder of this topic is structured as follows:

• Chapter 2 introduces the initially available Bio-jETI framework in de-

tail, before it deals with adding semantic awareness to workflow manage-
ment, especially concerning the support of workflow design by means of
constraint-based workflow development methods. In particular, the chap-
ter explains the extension of Bio-jETI with constraint-driven methods in
order to provide domain-specific support for (semi-) automatic workflow
design according to the principles of the loose programming paradigm.

• Chapters 3 - 6 illustrate Bio-jETI's application to the development of

bioinformatics workflows by means of four different application scenar-
ios from different thematic areas (phylogenetic analyses, PCR primer de-
sign, metabolic flux analysis, microarray data analysis) and with different
technical characteristics. For each of the four application scenarios, the
respective chapter comprises three major parts: