Introduction - User-Level Workflow Design: A Bioinformatics Perspective - page 22

1.2 Results

In the context described in the previous section, this topic focuses on workflow
management for the bioinformatics application domain, which is an interest-
ing “playground” for workflow methodologies for different reasons:

• Processing the large amounts of often heterogeneous data that is produced

by the life-science laboratories requires computational analysis procedures
(in bioinformatics jargon also called “in silico experiments”, referring to
and contrasting the traditional in vivo and in vitro experimentation in
biology). Rather than having a single piece of software ready for the job,
typically combinations of several resources and tools are required for per-
forming the possibly very complex, variant-rich analysis processes.

• The workflow approach to in silico experimentation enjoys great popular-

ity in the community, and accordingly a lot of research has been going on
in this field in recent years (cf., e.g., [92, 148]).

• In order to make tools and data easily available to researchers, the com-

munity has on the one hand adopted the use of (Web) Services for making
tools and data remotely accessible for the public, and on the other hand
many resources are open source and can deliberately be used. Thus, both
comprehensive data and service repositories are available and facilitate
the experimentation with realistic bioinformatics workflow scenarios.

The bioinformatics community is one of the most advanced with respect
to the emerging Semantic Web [40], actively working on providing domain-
specific meta-data for all kinds of resources [57]. Hence, the application
of semantics-based methods is largely facilitated.
As such, the bioinformatics application domain is particularly suited for
considering workflow management methodologies with a focus on variant-rich
scientific applications. Owing to the observation that especially with regard
to the workflow design phase, future developments must aim at lowering the
technical knowledge that is required for successfully combining services into
workflows (cf. Section 1.1.2), this topic addresses in particular the question
how workflow systems can offer better means for finding appropriate services,
for dealing with data formats, and for using the technical language of the
user rather than computer science terminology for communicating the user's
intents about his workflows. Therefore, this comprises two principal parts of
work, which are detailed in the following two sections:

•

1. The conception of a concrete software framework for user-level design of
workflows based on semantic techniques.
2. Comprehensive applications of the developed methods in the bioinformat-
ics domain in order to assess their capabilities and limitations.

In total, the systematic evaluation of the loose programming approach for
applicability in the bioinformatics domain steps into a new line of research on
user-level workflow design. It has great potential to be continued in different