Open Midnighter opened 7 years ago
Hi @Midnighter
I think we need to make sure have a coherent nomenclature, that is well documented:
boundaries
: all reactions that look like metabolite <->
despite of their upper and lower bound.
exchanges
: all boundaries
in the extracellular compartment.
sinks
: all boundaries
with lower_bound >= 0 and upper_bound > 0
. things going out
sources
: all boundaries
with lower_bound < 0 and upper_bound <= 0
. things going in
demands
: all sinks
in the extracellular compartment.
uptakes
: all sources
in the extracellular compartment.
The upper and lower bounds can be subject of discussion, but we need to get this sorted out nicely!
Regarding sinks
and sources
it has been defined in graph theory "a source vertex is a vertex with indegree zero, while a sink vertex is a vertex with outdegree zero".
More arguments in favor of using sinks and sources are their meaning: https://en.oxforddictionaries.com/definition/sink
https://en.oxforddictionaries.com/definition/source
While your definitions make sense to me I'd rather stick with conventions in the field. In Figure 7 [1]_,
.. [1] Thiele, Ines, and Bernhard Ø Palsson. “A Protocol for Generating a High-Quality Genome-Scale Metabolic Reconstruction.” Nature Protocols 5, no. 1 (January 2010): 93–121. doi:10.1038/nprot.2009.203.
I do agree with Joao, that in particular the literal wording of the defined term 'Sink' seems counterintuitive. A 'Pool' would seem to be the more appropriate terminology in my opinion. However, it may be possible that the original authors are referring something like this. Which says:
In all cases, each of the opposing terms (source or sink) may refer to the same object, depending on the perspective of the observer and the sign convention being used; there is no intrinsic difference between a source and a sink.
I'd like to highlight the following sentence as it might provide more clarity with respect to a biological example:
In biology, the schematic barrier in the figure could represent a cell membrane, and as a result, the two compartments could represent the inside and the outside of the cell. Generally speaking the point of observation would be outside the cell. Thus the cell would be termed a sink with respect to any flow of positive charges into it, and the cell would act as a source for any positive charges flowing out of it. Note that when considering the flow of negative charges, the definitions are reversed.
I'd prefer to keep the overall amount of definitions to a minimum.
Just found that a cobrapy model defines the property
boundaries
whereas the pathway predictor expectsexchanges
. We are thinking about implementing further methods/properties here (https://github.com/opencobra/cobrapy/pull/589) but we should make sure the right one is used by cameo (boundaries
includes sink and demand reactions).