fix: refine reactions, reversibility, and GPRs of Isocitrate Dehydrogenase metabolism

[haowang-bioinfo]

This PR improves GPRs of isocitrate dehydrogenase reactions, as proposed by #523:

• remove MAR04111, MAR04112, MAR04113, MAR04585, MAR04586, MAR04587, MAR04588, because as reported in Parallel/Redundant Versions Of Reactions #524:
  • MAR00710 is equivalent to MAR04111+MAR04585
  • MAR00712 is equivalent to MAR04113+MAR04587
  • MAR04588 is part of MAR03957
  • MAR03958 is equivalent to MAR04112+MAR04586
• change to reversible reactions for MAR00710, MAR00712, and MAR03958
• change GPRs to
  • ENSG00000138413: for MAR0710, MAR0712,
  • ENSG00000182054: for MAR03958
  • ENSG00000067829 and ENSG00000101365 and ENSG00000166411: for MAR03957
• associate Rhea IDs to MAR0710, MAR0712, MAR03957, MAR03958

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

[haowang-bioinfo]

@Devlin-Moyer please review the changes made to Isocitrate Dehydrogenase reactions based on your proposal.

[haowang-bioinfo]

@feiranl maybe you can double check this

[feiranl]

These commits are not fully documented in the #523 such as MAR03957 removal, and the proposed changes remain such as

The aconitase reaction (citrate <-> isocitrate) can be carried out in
one step:MAR04456
two steps: MAR04458 (citrate <-> cis-aconitate + H2O) and MAR04589 (cis-aconitate + H2O <-> isocitrate)

@haowang-bioinfo could you explain it a little bit? Actually, I am not a fan of removing the multi-step reactions, since it may affect the flux distribution. For example in the mentioned loop MAR04456: MAR04458 + MAR04589, what if citrate is not converted to isocitrate 100%, but it was converted to cis-aconitate and then it was converted to other metabolites, then by removing the MAR04458, we actually manipulate the flux distribution to fully to isocitrate.

[Devlin-Moyer]

all but two of the other reactions that cis-aconitate can participate in are transport reactions catalyzed by SLC25A1, and all of those reactions have an otherwise equivalent version that uses citrate or isocitrate in place of cis-aconitate, so removing the only sources of cis-aconitate aside from its exchange reaction would probably have no real impact on predicted/possible flux distributions

Reaction ID	Equation	Genes
MAR00125	citrate [c] + malate [m] ⇔ citrate [m] + malate [c]	SLC25A1
MAR02216	isocitrate [m] + malate [c] ⇔ isocitrate [c] + malate [m]	SLC25A1
MAR02420	cis-aconitate [m] + malate [c] ⇔ cis-aconitate [c] + malate [m]	SLC25A1
MAR02220	citrate [c] + succinate [m] ⇔ citrate [m] + succinate [c]	SLC25A1
MAR02413	isocitrate [m] + succinate [c] ⇔ isocitrate [c] + succinate [m]	SLC25A1
MAR02418	cis-aconitate [m] + succinate [c] ⇔ cis-aconitate [c] + succinate [m]	SLC25A1
MAR02405	AKG [c] + citrate [m] ⇔ AKG [m] + citrate [c]	SLC25A1
MAR02303	AKG [c] + isocitrate [m] ⇔ AKG [m] + isocitrate [c]	SLC25A1
MAR02416	AKG [c] + cis-aconitate [m] ⇔ AKG [m] + cis-aconitate [c]	SLC25A1
MAR02406	citrate [m] + oxalate [c] ⇔ citrate [c] + oxalate [m]	SLC25A1
MAR02414	isocitrate [m] + oxalate [c] ⇔ isocitrate [c] + oxalate [m]	SLC25A1
MAR02422	cis-aconitate [m] + oxalate [c] ⇔ cis-aconitate [c] + oxalate [m]	SLC25A1
MAR04971	PEP [c] + citrate [m] ⇔ PEP [m] + citrate [c]	SLC25A1
MAR06297	PEP [c] + isocitrate [m] ⇔ PEP [m] + isocitrate [c]	SLC25A1
MAR06298	PEP [c] + cis-aconitate [m] ⇔ PEP [m] + cis-aconitate [c]	SLC25A1
MAR06286	AKG [m] + H+ [c] + citrate [c] ⇒ AKG [c] + H+ [m] + citrate [m]	SLC25A1
MAR06289	AKG [m] + H+ [c] + isocitrate [c] ⇒ AKG [c] + H+ [m] + isocitrate [m]	SLC25A1
MAR06293	AKG [m] + H+ [c] + cis-aconitate [c] ⇒ AKG [c] + H+ [m] + cis-aconitate [m]	SLC25A1
MAR06287	H+ [c] + citrate [c] + succinate [m] ⇒ H+ [m] + citrate [m] + succinate [c]	SLC25A1
MAR06290	H+ [c] + isocitrate [c] + succinate [m] ⇒ H+ [m] + isocitrate [m] + succinate [c]	SLC25A1
MAR06294	H+ [c] + cis-aconitate [c] + succinate [m] ⇒ H+ [m] + cis-aconitate [m] + succinate [c]	SLC25A1
MAR06288	H+ [c] + citrate [c] + oxalate [m] ⇒ H+ [m] + citrate [m] + oxalate [c]	SLC25A1
MAR06292	H+ [c] + isocitrate [c] + oxalate [m] ⇒ H+ [m] + isocitrate [m] + oxalate [c]	SLC25A1
MAR06296	H+ [c] + cis-aconitate [c] + oxalate [m] ⇒ H+ [m] + cis-aconitate [m] + oxalate [c]	SLC25A1
MAR04964	H+ [m] + citrate [m] + malate [c] ⇒ H+ [c] + citrate [c] + malate [m]	SLC25A1
MAR06291	H+ [c] + isocitrate [c] + malate [m] ⇒ H+ [m] + isocitrate [m] + malate [c]	SLC25A1
MAR06295	H+ [c] + cis-aconitate [c] + malate [m] ⇒ H+ [m] + cis-aconitate [m] + malate [c]	SLC25A1
MAR04972	citrate [c] + isocitrate [m] ⇔ citrate [m] + isocitrate [c]	SLC25A1
MAR06276	cis-aconitate [c] + citrate [m] ⇔ cis-aconitate [m] + citrate [c]	SLC25A1
MAR06277	cis-aconitate [c] + isocitrate [m] ⇔ cis-aconitate [m] + isocitrate [c]	SLC25A1

the only reactions involving cis-aconitate not on this table are the two steps of the reaction catalyzed by aconitase mentioned above (MAR04458 and MAR04589), the exchange reaction (MAR11350), and symport of cis-aconitate and sodium across the cell membrane (MAR099287)

[haowang-bioinfo]

These commits are not fully documented in the #523 such as MAR03957 removal

these commits didn't remove MAR03957, and the removal here includes only 7 reactions that are: MAR04111, MAR04112, MAR04113, MAR04585, MAR04586, MAR04587, MAR04588 (this is stated in the first point of PR message)

@haowang-bioinfo could you explain it a little bit? Actually, I am not a fan of removing the multi-step reactions, since it may affect the flux distribution. For example in the mentioned loop MAR04456: MAR04458 + MAR04589, what if citrate is not converted to isocitrate 100%, but it was converted to cis-aconitate and then it was converted to other metabolites, then by removing the MAR04458, we actually manipulate the flux distribution to fully to isocitrate.

this PR didn't touch MAR04456, MAR04458, or MAR04589.

[haowang-bioinfo]

Regarding the case of MAR04456: MAR04458 + MAR04589, in general I'm kind of leaning toward the idea of @feiranl, i.e. retaining the high resolution version to capture all possible fluxes

with the very nice table prepared by @Devlin-Moyer, it seems that the low resolution option is simple and good enough, according to Occam's razor principle

here's one additional piece of evidence that reports mitochondrial TCA transporter Mtt1 that can relocate cis-aconitate from mitochondrion to cytosol

[haowang-bioinfo]

for resolving the duplication of MAR04456 vs. MAR04458 + MAR04589, how about remove MAR04456, this would:

1. be the minimum change to the model
2. without affecting listed transport reactions above
3. higher granularity in resolution supporting potential transport of cis-aconitate from [m] to [c]

what do you think? @Devlin-Moyer @feiranl

[Devlin-Moyer]

Sounds good to me

[haowang-bioinfo]

Sounds good to me

okay, implemented in 8e4f2d2

[feiranl]

I found after this PR, there are three metabolites MAM02662c, MAM02662m, MAM02662x (oxalosuccinate) that are not used in any rxn in the model. I checked this is due to removing rxns MAR04111,MAR04112,MAR04113,MAR04585,MAR04586,MAR04587,MAR04588. According to this PR, they were removed due to parallel or redundant rxns existed. Do you have any suggestions?

[haowang-bioinfo]

Good spot! you are absolutely right about the three orphan metabolites (MAM02662c, MAM02662m, MAM02662x) that currently do not involve into any reactions.

[haowang-bioinfo]

They should be removed while resolving the duplicated IDH rxns but somehow missed, and now fixed in #555.

How did you find this @feiranl? nice to include such check as part of GH action as well

[haowang-bioinfo]

nice to include such check as part of GH action as well

now implemented in #556