Feat/SBOterms

[BenjaSanchez]

Main improvements in this PR:

Following up on #106, an important model feature for memote is the existence of SBO terms for all components in the model. Here we have included SBO terms for all rxns and mets based on an automatic script that runs each time the model is updated with saveYeastModel.m, based on the criteria from addSBOterms.m:

• For metabolites:

SBO term	name	criterion	# in model
SBO:0000649	biomass	any metabolite name matching biomass, DNA, RNA, protein, carbohydrate or lipid, or ending with backbone or chain (pseudo-metabolites in SLIME reactions)	50 mets
SBO:0000247	simple chemical	the rest	2191 mets

• For reactions:

SBO term	name	criterion	# in model
SBO:0000627	exchange reaction	any reaction with only one metabolite involved, in the extracellular compartment	165 rxns
SBO:0000628	demand reaction	any reaction with only one metabolite involved, not in the extracellular compartment, and being produced	0 rxns
SBO:0000632	sink reaction	any reaction with only one metabolite involved, not in the extracellular compartment, and being consumed	3 rxns
SBO:0000629	biomass production	the reaction matching biomass pseudoreaction	1 rxn
SBO:0000630	ATP maintenance	the reaction matching non-growth associated maintenance reaction	1 rxn
SBO:0000395	encapsulating process	any reaction containing the words pseudoreaction or SLIME rxn	193 rxns
SBO:0000655	transport reaction	any reaction with at least one metabolite being moved from one compartment to another (same name but different compartment)	917 rxns
SBO:0000176	biochemical reaction	the rest	2240 rxns

@Midnighter @ChristianLieven opinions on this scheme?

I hereby confirm that I have:

• Tested my code with all requirements for running the model
• Selected devel as a target branch (top left drop-down menu)
• If needed, asked first in the Gitter chat room about this PR

[ChristianLieven]

Hi Ben, just a few comments on the current state of SBO testing in memote. In general, it is not fully developed yet. We're really specifically checking for a few node SBO terms and not yet for the presence of any children of those nodes.

• We don't test for biomass metabolites using SBO:0000649 (yet). There is only one function needs to identify biomass metabolites and here we currently look for biomass metabolites matching this regex ^biomass(_[a-zA-Z]+?)*?$ i.e. we look for metabolite IDs with the prefix biomass_. However, I believe this has to change in the future.

• As for metabolites we first test if there is generally any SBO term defined, and them more specifically we check how many metabolites are defined as for the presence of SBO:0000247 i.e. 'simple chemical'. Again, this will have to change in the future, when we start testing for this term and all of its children.

• For transport reactions, again, we're only looking at the parent node SBO:0000185 not at any of its children 'SBO:0000588', 'SBO:0000587', 'SBO:0000655', 'SBO:0000654',
  'SBO:0000660', 'SBO:0000659', 'SBO:0000657', and 'SBO:0000658'

• ATP maintenance and encapsulating process are not tested for either, but I think the former is definitely a thing we should do too! I'll add that to the issue about our SBO overhaul (Improve SBO term handling opencobra/memote#469).

Thanks again for giving us this feedback Ben! It's much appreciated!

[BenjaSanchez]

@ChristianLieven thanks a lot for the comments!

We don't test for biomass metabolites using SBO:0000649 (yet).

The biomass reaction is at least recognized, so for now this is not a big concern for us (just 98% instead of a 100% in the "Metabolites without SBO:0000247" test)

For transport reactions, again, we're only looking at the parent node SBO:0000185

The memote report states:

'SBO:0000185', 'SBO:0000588', 'SBO:0000587', 'SBO:0000655', 'SBO:0000654', 'SBO:0000660', 'SBO:0000659', 'SBO:0000657', and 'SBO:0000658' represent the terms 'transport reaction' and 'translocation reaction', in addition to their children (more specific transport reaction labels). Every transport reaction that is not a pure metabolic or boundary reaction should be annotated with one of these terms.

So I guess this will also change in the future to account for all children?

For the time being this PR will be merged, and after memote finishes developing SBO testing we will asses if any of our criteria should change. Issue #106 will be adapted accordingly.

[hongzhonglu]

@BenjaSanchez
As for the SBO term, I find that r_0226, r_0438, r_0439... belong to SBO:0000655 (transport type), however, these reactions are not belong to transport reaction. Can you check it?

[BenjaSanchez]

@hongzhonglu those 3 rxns are 3 of the oxidative phosphorylation complexes, and they do pump protons through the mitochondrial membrane (in one direction or the other), so from the SBO perspective they can be categorized as transports. If you think they are better suited to another ontology feel free to open an issue for further discussion

[hongzhonglu]

@BenjaSanchez OK! Thanks.