Fix: Make all non-transport/exchange pyrophosphate reactions irreversible

[Devlin-Moyer]

Main improvements in this PR:

Makes all reversible non-transport/exchange reactions that involve pyrophosphate (MAM02759e/c/m/x/n/r) irreversible in the direction of pyrophosphate production, as discussed in #527. In cases where pyrophosphate is currently a “reactant,” products and reactants are swapped to ensure no reactions are constrained to only have fluxes <= 0.

I hereby confirm that I have:

• Tested my code on my own computer for running the model
• Selected develop as a target branch
• Any removed reactions and metabolites have been moved to the corresponding deprecated identifier lists

[Devlin-Moyer]

Oh dang; I used Cobrapy's built-in function for saving models in YAML format, but it appears to have not used the format your automated format checker was expecting

[haowang-bioinfo]

looks good

I'm afraid the effect of these changes needs to be checked with verification and essentiality tests that are under construction, will come back as soon as possible

[haowang-bioinfo]

Oh dang; I used Cobrapy's built-in function for saving models in YAML format, but it appears to have not used the format your automated format checker was expecting

actually is considering to fix the compatibility to cobrapy, any ideas in improving this?

[Devlin-Moyer]

Cobrapy uses ruamel.yaml to generate their YAML files. I'm not certain if these are all of the differences, but:

• Cobrapy's YAML files don't have the header section that yours does
• Cobrapy's YAML files don't quote all character/text/non-numeric fields by default, while yours do (without the quotes, your automated validation bot kept interpreting the "formula" entires for nitrous oxide (NO) metabolites as if they were supposed to be True/False and telling me it couldn't interpret "NO" as either "True" nor "False")
• Cobrapy's YAML files use two spaces for each level of indentation, but yours have 2 spaces for the first level, 4 spaces for the second, 2 for the third, etc.
• Cobrapy's YAML files have a separate "id" section (for the model id) at the "root" level (i.e. the same level as the metabolites, reactions, genes, and compartments)
• Cobrapy's YAML files seem to wrap/split lines that are over 80 characters across multiple lines, which only seemed to affect metabolites with long names and reactions with long gene_reaction_rules (also your automated validation bot never flagged that as an issue)

I suspect that the biggest issues are the contents of the header and the quotes around metabolite formulas; I think fixing the header problem would require changes on Cobrapy's end, because, as far as I can tell, most of the information you have in the header of your YAML file isn't in a Cobrapy YAML file at all. The other things seem like things you could probably tweak with your YAML validation thing to accommodate, but I'm not entirely sure how it works.

Also it may be worth noting that Cobrapy had no problem reading in Human-GEM.yml, as far as I could tell

[mihai-sysbio]

While I realise it is more cumbersome, SBML is a more robust file format for using the model across different tools/platforms.

[haowang-bioinfo]

it may be worth noting that Cobrapy had no problem reading in Human-GEM.yml, as far as I could tell

yes, the reading was fixed early on; now probably is time to resolve writing side

Cobrapy's YAML files don't have the header section that yours does
Cobrapy's YAML files don't quote all character/text/non-numeric fields by default, while yours do (without the quotes, your automated validation bot kept interpreting the "formula" entires for nitrous oxide (NO) metabolites as if they were supposed to be True/False and telling me it couldn't interpret "NO" as either "True" nor "False")
Cobrapy's YAML files use two spaces for each level of indentation, but yours have 2 spaces for the first level, 4 spaces for the second, 2 for the third, etc.
Cobrapy's YAML files have a separate "id" section (for the model id) at the "root" level (i.e. the same level as the metabolites, reactions, genes, and compartments)
Cobrapy's YAML files seem to wrap/split lines that are over 80 characters across multiple lines, which only seemed to affect metabolites with long names and reactions with long gene_reaction_rules (also your automated validation bot never flagged that as an issue)

thanks for listing the details

I suspect that the biggest issues are the contents of the header and the quotes around metabolite formulas; I think fixing the header problem would require changes on Cobrapy's end, because, as far as I can tell, most of the information you have in the header of your YAML file isn't in a Cobrapy YAML file at all. The other things seem like things you could probably tweak with your YAML validation thing to accommodate, but I'm not entirely sure how it works.

the quote issue is technically solvable, while header section actually is the tricky part

[haowang-bioinfo]

I realise it is more cumbersome, SBML is a more robust file format for using the model across different tools/platforms.

yes SBML is a mess

[mihai-sysbio]

would require changes on Cobrapy's end

This might be the better solution in more than one case.

[haowang-bioinfo]

I'm afraid the effect of these changes needs to be checked with verification and essentiality tests that are under construction, will come back as soon as possible

@Devlin-Moyer would you be able to merge the latest develop branch into this PR, or start a new PR from the latest develop to add the commits (then close this PR). The idea is to make sure that these changes won't break the model

[haowang-bioinfo]

very nice - significant progress in improving thermodynamic constrains!