FAD/Oxygen-Peroxide Duplicate Reactions

[Devlin-Moyer]

Problems

Similarly to #607, I found a number of pairs of reactions that are largely identical except one reduces FAD to FADH₂ and the other reduces O₂ to H₂O₂:

Reactant	FAD-dependent ID	O2-dependent ID
25(S)THCA-CoA	MAR01638	MAR01639
linoleoyl-CoA	MAR02804	MAR03301
cis,cis-palmito-7,10-dienoyl-CoA	MAR02808	MAR03306
cis,cis-myristo-5,8-dienoyl-CoA	MAR02810	MAR03311
4-cis-decenoyl-CoA	MAR02818	MAR04966
palmitoleoyl-CoA	MAR03276	MAR03364
(7Z)-tetradecenoyl-CoA	MAR03289	MAR03369
(5Z)-dodecenoyl-CoA	MAR03295	none

There's also this strange case:
MAR00059: 3alpha,7alpha-dihydroxy-5beta-cholest-24-enoyl-CoA [x] + FADH2 [x] + 0.5 O2 [x] ⇒ (24R,25R)3alpha,7alpha,24-trihydroxy-5beta-cholestanoyl-CoA [x] + FAD [x], GPR: ACOX2

Background/Rationale

ACOX2 irreversibly oxidizes its substrates (see Uniprot and this paper), so MAR00059 definitely doesn't represent a reaction catalyzed by ACOX2 (see #634 for details and examples of reactions that ACOX2 should be associated with). However, MAR00059 is very similar to MAR01706:

MAR00059: 3alpha,7alpha-dihydroxy-5beta-cholest-24-enoyl-CoA [x] + FADH2 [x] + 0.5 O2 [x] ⇒ (24R,25R)3alpha,7alpha,24-trihydroxy-5beta-cholestanoyl-CoA [x] + FAD [x], GPR: ACOX2
MAR01706: 3alpha,7alpha-dihydroxy-5beta-cholest-24-enoyl-CoA [x] + H2O [x] ⇒ (24R,25R)3alpha,7alpha,24-trihydroxy-5beta-cholestanoyl-CoA [x], GPR: ACAA1 and EHHADH and HSD17B4

According to this paper about bile acid biosynthesis, MAR01706 already accurately represents this step in the pathway and the catalytic activities of EHHADH and HSD17B4 (also see their Uniprot pages), so MAR00059 should just be removed for being a less-accurate version of MAR01706. Relatedly, ACAA1 is a 3-ketoacyl thiolase, and MAR01706 doesn't involve any thiolysis, so it should be removed from the GPR of MAR01706.

Most similar reactions (i.e. the ~100 other reactions associated with ACOX1, ACOX2, or ACOX3 that involve peroxisomal metabolites) reduce O₂ to H₂O₂ and not FAD to FADH₂, so it would be easier/simpler to remove the FAD-dependent versions and keep the O₂-dependent versions. However, that would be inconsistent with the changes and rationale in #607, where we kept the FAD-dependent versions of the reactions to explicitly account for its role in the transfer of electrons from various substrates to ubiquinone. Also, FAD cofactors are not covalently attached to most human enzymes (source), FAD cannot be biosynthesized in human peroxisomes (source), and SLC25A17 (ENSG00000100372) exchanges cytosolic FAD for peroxisomal FMN in humans (same source as before). Keeping the FAD-dependent members of each of these pairs, replacing O₂ and H₂O₂ with FAD and FADH₂ in all similar reactions, and adding a separate reaction for reduction of O₂ to H₂O₂ by FADH₂ would be a lot of work, but it would be more realistic (and consistent with #607).

Proposed Changes

• Remove MAR00059 for being a less accurate version of MAR01706
• Change the GPR of MAR01706 to ENSG00000113790 or ENSG00000133835 and add PMID:19357427 to its references
• Remove MAR01639 for being redundant with MAR01638
• Remove MAR03301 for being redundant with MAR02804
• Remove MAR03306 for being redundant with MAR02808
• Remove MAR03311 for being redundant with MAR02810
• Remove MAR04966 for being redundant with MAR02818
• Remove MAR03364 for being redundant with MAR03276
• Remove MAR03369 for being redundant with MAR03289
• Replace MAM02630x (O₂) with MAM01802x (FAD) and MAM02041x (H₂O₂) with MAM01803x (FADH₂) in MAR00442, MAR00759, MAR00777, MAR00800, MAR00833, MAR00872, MAR00952, MAR00968, MAR00974, MAR00975, MAR00977, MAR00978, MAR00982, MAR01001, MAR01005, MAR01018, MAR01162, MAR01205, MAR01254, MAR01271, MAR01291, MAR01296, MAR01303, MAR01600, MAR01704, MAR02408, MAR02510, MAR03053, MAR03054, MAR03055, MAR03056, MAR03057, MAR03062, MAR03066, MAR03070, MAR03074, MAR03078, MAR03082, MAR03086, MAR03090, MAR03094, MAR03098, MAR03102, MAR03326, MAR03330, MAR03334, MAR03338, MAR03342, MAR03346, MAR03350, MAR03356, MAR03360, MAR03459, MAR03466, MAR03470, MAR03498, MAR03506, MAR03512, MAR03849, MAR03984, MAR04001, MAR04650, MAR04789, MAR04981, MAR05025, MAR05026, MAR05028, MAR05097, MAR05100, MAR05108, MAR05119, MAR05175, MAR05178, MAR05180, MAR05182, MAR05184, MAR05189, MAR05196, MAR05300, MAR05327, MAR05331, MAR05350, MAR05366, MAR05372, MAR05374, MAR05377, MAR05382, MAR05421, MAR05428, MAR05431, MAR06284, MAR06285, MAR06322, MAR06344, MAR06345, MAR06933, MAR06936, MAR06943, MAR06946, MAR06949, MAR06952, MAR06959, MAR06962, MAR06964, MAR07641, MAR07647, MAR07704, MAR07706, MAR08017, MAR08021, MAR08415, MAR08689, MAR08690, MAR08755, MAR11529, MAR11534, MAR11539, MAR12372, MAR12808, and MAR20066 (every reaction that consumes O₂ and produces H₂O₂ in the peroxisome except the ones that already have FAD-dependent versions)
• Create new reaction to represent reduction of O₂ by FADH₂: MAM02630x + MAM01803x -> MAM02041x + MAM01802x, GPR: ENSG00000161533 or ENSG00000168306 or ENSG00000087008 or ENSG00000110887 or ENSG00000203797 or ENSG00000007171 or ENSG00000148832 or ENSG00000179761 or ENSG00000158125 (every enzyme mentioned here except ACOXL, which the paper says is not known to be catalytically active, UOX, which is a pseudogene in humans, and HAO1 and HAO2, which use FMN and not FAD according to this paper)

[haowang-bioinfo]

nice and thorough investigation, and a large chunk of the network would be affected by the proposed changes

[feiranl]

Please manually test all metabolic tasks can be filled before making the changes.

[Devlin-Moyer]

I don't know what that means; which metabolic tasks are we talking about?

[feiranl]

The task list can be found here, and @haowang-bioinfo maybe you can guide how to test those tasks. I only find the RAVEN function checkTasks, not sure wether this is correct.

taskStruct = parseTaskList('../data/metabolicTasks_Essential.txt');
[taskReport, essentialRxnMat, ~, essentialFluxes] = checkTasks(bModel,[],true,false,true,taskStruct);

[JonathanRob]

When running checkTasks in this case, you can set the getEssential input to False, since you don’t need to get the essential reactions, just test if the tasks pass. So the second line could just be:

checkTasks(bModel,[],true,false,false,taskStruct);

This will greatly decrease run time, since the essentiality part takes a while.

[haowang-bioinfo]

I don't know what that means; which metabolic tasks are we talking about?

feel free to start a PR which can be made directly from a branch within Human-GEM, rather than from a forked repo (then we can check out the tasks). Is this clear? @Devlin-Moyer

[Devlin-Moyer]

yea I can do that once #624 is merged; I've been holding off on opening any other PRs that add new reactions until #624 is merged since that adds new reactions, and having multiple PRs that add new reactions open at the same time seems like it'd set us up for some confusing/annoying/complicated merge conflicts and/or distinct reactions with the same IDs

[Devlin-Moyer]

Addressed by #670