FiatFlux-P - User-Level Workflow Design: A Bioinformatics Perspective - page 109

result has to be provided. Both these variations have been extensively used
in the Laboratory of Chemical Biotechnology, TU Dortmund, for the analysis
of various experimental data.
As graphical presentation of metabolic flux distributions greatly facilitates
their interpretation, the FiatFlux code has furthermore been extended by
functionality for exporting the calculated fluxes in a format that can be used
for visualizing the calculated metabolic flux distributions using the OMIX
software [88], an editor for drawing metabolic reaction networks. More pre-
cisely, the reaction rates together with specific reaction identifiers are ex-
ported to a .csv file, which can be saved from within the workflow just as the
ratio and netto result files.
This format can later on easily be interpreted by an OVL (OMIX Visu-
alization Language) script in order to equip default network diagrams with
markups according to the obtained results. The customized metabolic flux
chartscanthenbeexportedbyOMIX into different bitmap and vector
graphic formats such as .png, .jpg and .svg. Within the FiatFlux-P project,
ready-to-use OMIX network diagrams for a number of frequently analyzed
model organisms are provided, along with an OVL script offering two dif-
ferent markup variants, one for the visualization of a single result data set,
where the line width of the reaction arrows is adjusted to the specific flux,
and another for the visualization of multiple result data sets, where the values
of the reaction rates are assigned to the arrows representing the respective
reaction.
5.3 Constraint-Driven Design of FiatFlux-P Workflows
In the previous examples, the tools and algorithms that were involved were
mostly more or less common, third-party services, and thus it was indicated to
automatically retrieve the basic domain information from third-party knowl-
edge repositories, too. FiatFlux-P, in contrast, is a very specific application,
tailored to the development of workflows for metabolic flux analysis with
a particular variant of the FiatFlux software. Hence the EDAM ontology
does not comprise the suitable terminology for this application. However,
FiatFlux-P uses a comparatively small set of services and comprises only a
small number of different data types, so that setting up a clear and unam-
biguous domain model is in fact straightforward. In particular, the service
and type taxonomies have been defined manually and closely tailored to the
specific application examples, providing exactly the level of generalization
and refinement that is required for this particular application. The corre-
sponding domain model for FiatFlux-P is presented in the following, before
an exemplary workflow composition problem is discussed.