Processes 2018, 6,39
Reaction Names (optional): The reaction names on the SVG representation are necessary if you
want to write down the flux values on the SVG image. The ID must be identical to the reaction
name (without the counter and additional characters). It must be in a SVG tspan element of a text
element. The SVG tspan element of the reaction name is not allowed to carry any other text than the
reaction name.
Placing the Flux text (optional): It is also possible to write down the flux distribution as text on
the image. To enable this the network file requires a text element saying: “place_here” (Figure 1a).
This text element fulfils the same constraints as the reaction names mentioned above. At the position
of this text element, the program will write down the flux distribution (Figure 1b, the text “place_here”
in (a) is replaced by “Example Flux: 0.22 T6 ...” in (b)).
4. Discussion and Conclusions
As mentioned in the introduction, many software already exists to automatically generate flux
maps in metabolic network (at the genome-scale level or not). Usually they are able to draw the entire
metabolic network and the pathways of interest with possibly several layouts and zoom functions.
However, there is, among biochemists, a long tradition of metabolic network representation
with simple arrows as exemplified in biochemistry textbooks (and in Figures 1 and 5) with particular
disposition in space. A simple vertical line with a circle underneath irreversibly evokes glycolysis
and the Krebs cycle for a biochemist. It is important to keep this implicit knowledge to facilitate the
interpretation of metabolic results or hypotheses. It is undoubtedly the reason why experimental
biochemists often draw their own metabolic network representation by hand and report flows also by
hand with values and/or colours. This process is tiresome and time-consuming and was automatized
with FluxVisualizer. This is why FluxVisualizer does not draw a representation of metabolic networks
but starts from the drawings of the biochemist himself with a minimum of constraints.
It is thus difficult to compare FluxVisualizer with other software mainly dealing (among other
functionalities) with network representation. These software can be used in synergy with FluxVisualizer
in providing an SVG representation FluxVisualizer will exploit. This is done on Figure 6 using
MetExplore, which is rather easy to use and can export an SVG image, in this case the same metabolic
network as in Figure 1a. It must be noted however, that the presence of the nodes, rectangles and
circles representing the reactions and metabolites unnecessarily clutters the larger schemes and makes
them less readable. Furthermore, it is necessary to re-arrange spatially this diagram to reproduce
the biochemist’s layout of Figure 1a. It necessitates a long and tedious work. Undeniably, SVG
representation is not (yet) a standard among experimental biologists and this requirement may turn
off biologists from FluxVisualizer use, contradicting the purpose of FluxVisualizer to be tailored to
biochemists. This is a difficulty for a biochemist to a friendly use of FluxVisualizer. There is another
way to take this deterrent step forward. Very often, biologists use LibreOffice (https://fr.libreoffice.org)
(or Microsoft Office (https://products.office.com)) suite to draw their diagrams of metabolic networks.
It is easy to save them in the pdf format which can be read by Inkscape, for instance, and be converted
to the SVG format. This was done in the case of Figure 1 which, initially, was a PowerPoint file. It is
then necessary to check the arrow’s ID in order for them to match exactly the name of the reaction.
The last solution is to build directly the diagram with Inkscape, for instance, which is not so difficult to
manage. This was done for Figure 5.
Another difficulty in visualizing the FluxVisualizer results are the overlaps of the flux values with
the rest of the initial image (see Figure 1b). These unescapable overlaps, are not too troublesome for
small metabolic networks for which the attribution of a flux value to an enzymatic step remains clear
but it is a limit in the size of the network. These overlaps cannot be avoided in an automatic process
but their effects can be diminished in playing with the different options of positioning the flux values
offered in FluxVisualizer.
Although there is, in principle, no limit in the size of the metabolic network that FluxVisualizer
can deal with (genome-scale network should be, in principle, handled) the increasing number of
150
Reaction Names (optional): The reaction names on the SVG representation are necessary if you
want to write down the flux values on the SVG image. The ID must be identical to the reaction
name (without the counter and additional characters). It must be in a SVG tspan element of a text
element. The SVG tspan element of the reaction name is not allowed to carry any other text than the
reaction name.
Placing the Flux text (optional): It is also possible to write down the flux distribution as text on
the image. To enable this the network file requires a text element saying: “place_here” (Figure 1a).
This text element fulfils the same constraints as the reaction names mentioned above. At the position
of this text element, the program will write down the flux distribution (Figure 1b, the text “place_here”
in (a) is replaced by “Example Flux: 0.22 T6 ...” in (b)).
4. Discussion and Conclusions
As mentioned in the introduction, many software already exists to automatically generate flux
maps in metabolic network (at the genome-scale level or not). Usually they are able to draw the entire
metabolic network and the pathways of interest with possibly several layouts and zoom functions.
However, there is, among biochemists, a long tradition of metabolic network representation
with simple arrows as exemplified in biochemistry textbooks (and in Figures 1 and 5) with particular
disposition in space. A simple vertical line with a circle underneath irreversibly evokes glycolysis
and the Krebs cycle for a biochemist. It is important to keep this implicit knowledge to facilitate the
interpretation of metabolic results or hypotheses. It is undoubtedly the reason why experimental
biochemists often draw their own metabolic network representation by hand and report flows also by
hand with values and/or colours. This process is tiresome and time-consuming and was automatized
with FluxVisualizer. This is why FluxVisualizer does not draw a representation of metabolic networks
but starts from the drawings of the biochemist himself with a minimum of constraints.
It is thus difficult to compare FluxVisualizer with other software mainly dealing (among other
functionalities) with network representation. These software can be used in synergy with FluxVisualizer
in providing an SVG representation FluxVisualizer will exploit. This is done on Figure 6 using
MetExplore, which is rather easy to use and can export an SVG image, in this case the same metabolic
network as in Figure 1a. It must be noted however, that the presence of the nodes, rectangles and
circles representing the reactions and metabolites unnecessarily clutters the larger schemes and makes
them less readable. Furthermore, it is necessary to re-arrange spatially this diagram to reproduce
the biochemist’s layout of Figure 1a. It necessitates a long and tedious work. Undeniably, SVG
representation is not (yet) a standard among experimental biologists and this requirement may turn
off biologists from FluxVisualizer use, contradicting the purpose of FluxVisualizer to be tailored to
biochemists. This is a difficulty for a biochemist to a friendly use of FluxVisualizer. There is another
way to take this deterrent step forward. Very often, biologists use LibreOffice (https://fr.libreoffice.org)
(or Microsoft Office (https://products.office.com)) suite to draw their diagrams of metabolic networks.
It is easy to save them in the pdf format which can be read by Inkscape, for instance, and be converted
to the SVG format. This was done in the case of Figure 1 which, initially, was a PowerPoint file. It is
then necessary to check the arrow’s ID in order for them to match exactly the name of the reaction.
The last solution is to build directly the diagram with Inkscape, for instance, which is not so difficult to
manage. This was done for Figure 5.
Another difficulty in visualizing the FluxVisualizer results are the overlaps of the flux values with
the rest of the initial image (see Figure 1b). These unescapable overlaps, are not too troublesome for
small metabolic networks for which the attribution of a flux value to an enzymatic step remains clear
but it is a limit in the size of the network. These overlaps cannot be avoided in an automatic process
but their effects can be diminished in playing with the different options of positioning the flux values
offered in FluxVisualizer.
Although there is, in principle, no limit in the size of the metabolic network that FluxVisualizer
can deal with (genome-scale network should be, in principle, handled) the increasing number of
150