Menu
▾
▴
#325 Mention using variable stoichiometries
closed
nobody
5
2017-12-06
2016-02-20
No
People often balk at the notion of variable stoichiometries, but the following search reveals there are models out there that use it:
https://www.google.co.uk/search?q=variable+stoichiometry+model
So, I propose that in the section on reactions, where the stoichiometry attribute is discussed, there should be a sentence to the effect that the SBML scheme permits variable stoichiometries, and then have a couple of references to models that use variable stoichiometries.
This makes sense given we have had stocihiometryMath and now have a different mechanism but essentially the same thing.
Here is the current discussion on variable stoichiometries. First, from the section on the 'stoichiometry' attribute:
"If the species reference's \token{constant} attribute
has the value \val{false}, the species reference's value may be
overridden by an \InitialAssignment or changed by \AssignmentRule
or \AlgebraicRule, and in addition, for simulation time $t > 0$,
it may also be changed by a \RateRule or \Event. (However,
some of these constructs are mutually exclusive; see
\sec{sec:rules} and \sec{sec:events}.)"
Next, from the section of the 'constant' attribute:
"A value of \val{false} means the stoichiometry can be changed by
other SBML constructs such as rules (see \sec{sec:rules}),
as described above in the section on the \token{stoichiometry}
attribute."
The only thing missing from your proposal would be pointing out specific models that use variable stoichiometry, or a general defense of why we allow this at all. Is it worth adding something to this effect? Or is the existing discussion adequate?
I think, refering to such models is a good idea. In this context, it would also be important to quote ME models that are currently not available in SBML formats, but highly depend on variable stoichiometries. See http://www.ncbi.nlm.nih.gov/pubmed/24084808 and subsequent papers.
I am adding a paragraph about variable stoichiometry models, and have cited Andreas's paper, but it would be nice to have a couple others. I'm unfamiliar with this subfield myself--would someone be willing to look up a couple other example papers (ideally not all from the same lab) and post links? I can work them into the text if so. Thanks!
Here's some draft text:
Variable stoichiometries are not often encountered in systems biology modeling, but there are subfields where the concept is vital to their models. Whole-cell growth models are one such example~\citep{obrien:2013}, as are [REF AND REF]. Variable stoichiometries are of critical importance to this type of modeling, and explains why the feature is included in SBML.
I could pick some random articles from https://scholar.google.com/scholar?hl=en&q=variable+stoichiometry+model&btnG=&as_sdt=1%2C48&as_sdtp= but I'm not confident enough in my understanding of the fields in question to pick the 'right' ones from there. I think a couple options beyond Andreas's will work.
Hello,
I am not entirely sure what the request was that we should vote on/discuss here?
Is it to discuss the notion of variable stoichiometry, or is it that someone suggested to add examples to the spec, or ...?
Side note:
I think that if we start adding examples for the use of an SBML concept in a model, then we should
(1) link to an SBML file rather than to a paper (BIOMID?)
(2) do this for ... all SBML concepts (and I think this would be a great idea, but an additional resource probably?)
Thanks for clarifying the issue again :)
Dagmar
Yes, at this point the proposal is to add the paragraph:
"Variable stoichiometries are not often encountered in systems biology modeling, but there are subfields where the concept is vital to their models. Whole-cell growth models are one such example~\citep{obrien:2013}, as are [REF AND REF]. Variable stoichiometries are of critical importance to this type of modeling, and explains why the feature is included in SBML."
but properly fill it in with a couple more references.
If there are any biomodels (either curated or non-curated) that have variable stoichiometry, I'd be happy to link to them; I'm not sure if any exist.
Thanks Lucian.
I am still a bit uncomfortable with having this extra explanation/justification for this specific SBML construct. I do not understand why we need it here (but not for the others).
Therefore I tend to disagree with the change.
Hi Dagmar,
In contrast to many other SBML constructs, variable stoichiometries reflect a concept that seems to be abstract and unnecessary at the first glance. As stated above, additional explaination is needed to show why this is important, relevant, and useful for the domain of modeling. Rather than adding a justification to every SBML concept, I propose to limit this to cases where it is really needed (either because it was requested or because we observe the lack of clear communication of the concept). I think it is better to cite publications that establish concepts in computational modeling rather than SBML files, because (as you know) many complex and highly advanced models might not even be available in SBML format yet.
Cheers
Andreas
SBML is a tool and methodology neutral encoding format and therefore references to specific (selected) implementations of any modelling methodology should be avoided. A specification should not be a document on how to model (implied or otherwise).
So my vote is no.
That's one editor (Andreas) for the change, and two against (Brett and Dagmar). Sarah and Nicolas, can you both chime in? We'll need both of your opinions on this one.
It might also be helpful if Mike said a little more about why he thought this would be an appropriate addition.
Andreas explained it well. The point of this was just to explain why variable stoichiometries are a feature. It was motivated by the fact that I have faced questions and complaints about the need for variable stoichiometries, and so it seemed that saying something in the spec would help. There are other places in the spec where we mention why other feature SBML exist, so this did not seem unusual to me. If people don't feel it's needed, then it's okay with me to leave it out
[sorry for the silence. I did comment on this one, but must have closed the window without clicking "post". I even read the papers listed by Lucian in order to see which ones are actually deading with variable stoichiometries in Systems Biology]
I am definitively for adding explanations. Variable stochiometry are useful and used. We removed the construct, and therefore should explain how to replace it.
I don't think anyone is against explaining how variable stoichiometries now work. The question was whether we should add a justification as to why we have variable stoichiometries at all. Such an explanation could have been added to the old 'stoichiometryMath' section as well as the new 'variable stoichiometry' section.
I dont think we have to 'justify' why things are in SBML at this stage. Things are there because people wanted/needed/used them. Maybe a sentence that just states:
Real world examples of variable stoichiometry in models are available {REF}.
This sort of proves the point that the concept is needed without sounding like we are apologising for it being there; or feeling that we have to justify it.
Sarah's compromise seems reasonable to me. Instead of a new paragraph, then, we would simply add the single sentence to the end of the last paragraph in the current 'stoichiometry attribute' section (p. 71 in the current draft spec):
"For examples of models with variable stoichiometries, see \citep{obrien:2013}, \citep{REF}, and \citep{REF}."
(Still need some suggestions for the other two REFs, or the go-ahead to pick some myself.)
I'll assume Sarah votes 'yes' to this--can the rest of you chime in?
Update summary of the above: Three editors voted against the 'full' version of the 'variable stoichiometry justification' paragraph proposed above, making the result ambiguous. Sarah then proposed a new version, which nobody besides her has commented on as of yet. Editors, please comment!
I can live with most of Sarah's sentence but why not simply let the reference be: google "models variable stoichiometry"? At least you rreferences won't be out of date as soon as the specification is published, it avoids makeing an (arbitrary) selection and you can hyperref it.
That would work for me--use the link
https://scholar.google.com/scholar?q=model+variable+stoichiometry
That works for me.
Usually, scientific findings are never out of date. Additional articles might appear, but this doesn't render previous ones invalid. But anyways, I support Sarah's compromise. For the reference we could cite some paper of particular importance and then still refer to the Google search if needed.
With three votes in favor of Sarah's revised sentence, I'm calling this issue 'pending'. The final version of the sentence I put in the spec is:
"Real world examples of variable stoichiometry in models include genome-scale models (O’Brien et al., 2013), and can generally be found in searches such as https://scholar.google.com/scholar?q=model+variable+stoichiometry."
We can certainly still tweak the sentence if we want.
Otherwise, it will be included as part of the next releases of SBML L3v1r2 and L3v2, and is included in the errata list at http://sbml.org/Documents/Specifications/SBML_Level_3/Version_1/Core/Confirmed_issues_in_the_Level_3_Version_1_Core_Specification
The above sentence is, in general, not completely clear. For example, in the FBC v2 strict interpretation, which covers the majority of current GSR's, the SpeciesReference must be flagged as constant. Where variable stoichiometry is relevant is in the specific type of GSR model described in the reference.
Does arbitrary cherry picking examples really address the problem? For me, no.
Last edit: Brett Olivier 2016-06-16
That makes sense! What about adding 'certain':
"Real world examples of variable stoichiometry in models include certain genome-scale models (O’Brien et al., 2013), and can generally be found in searches such as https://scholar.google.com/scholar?q=model+variable+stoichiometry."
Or we can just go with the search:
"Real world examples of variable stoichiometry in models can be found in the literature; searches such as https://scholar.google.com/scholar?q=model+variable+stoichiometry will reveal several such."