Duplicate Reactions For Alpha-Oxidation of Phytanoyl-CoA

[Devlin-Moyer]

Current behavior:

There are currently two parallel representations of the peroxisomal alpha-oxidation of phytanoyl-CoA to pristanic acid:

ID	Reaction	Genes
MAR03484	AKG + O2 + phytanoyl-CoA --> 2-hydroxyphytanoyl-CoA + CO2 + succinate	PHYH
MAR03486	2-hydroxyphytanoyl-CoA --> 2(S)-pristanal + formyl-CoA	HACL1 or KRTAP11-1
MAR03386	2(S)-pristanal + H2O + NAD+ <=> 2 H+ + NADH + pristanic acid	PHYH
MAR05186	2-hydroxyphytanoyl-CoA + H2O + NADP+ --> formyl-CoA + 2 H+ + NADPH + pristanic acid	ALDH3A2 and HACL1
MAR05187	2-hydroxyphytanoyl-CoA + H2O + NAD+ --> formyl-CoA + 2 H+ + NADH + pristanic acid	ALDH3A2 and HACL1

Expected behavior:

• MAR05187 is redundant with MAR03486 + MAR03386, so I think it should be removed.
• Uniprot seems to think that ALDH3A2 is NAD⁺-dependent, not NADP⁺-dependent, so I think MAR05186 is just wrong and should also be removed.
• I can find no evidence that KRTAP11-1 (ENSG00000182591) has any enzymatic activity at all, much less for the specific reaction that MAR03486 represents, so I think it should be removed from the GPR of MAR03486 and from Human-GEM entirely (MAR03486 is the only reaction it is currently associated with).
• The GPR of MAR03386 should be ALDH3A2, not PHYH (ENSG00000107537); PHYH only catalyzes 2-hydroxylation of various acyl-CoA esters, i.e. the reaction represented in MAR03484 -- ALDH3A2 is responsible for the oxidation of aldehydes to carboxylic acids (see Uniprot).

Proposed changes:

• Remove MAR05187 for being a duplicate of MAR03486 + MAR03386
• Remove MAR05186 for being a misrepresentation of the catalytic activity of ALDH3A2
• Change the GPR of MAR03386 to ENSG00000072210
• Remove ENSG00000182591 from the GPR of MAR03486 and everywhere else

[haowang-bioinfo]

very interesting topic - before going into the detail, I'd like to ask why are you looking into alpha-oxidation phytanoyl-CoA, or actually also aiming to systematically formulate entire beta oxidation? @Devlin-Moyer

[Devlin-Moyer]

I'm working on a new semi-automated model curation tool that identifies reactions that form loops capable of sustaining infinite fluxes, and one such loop was phytanic acid [c] -> phytanoyl-CoA [c] -> phytanoyl-CoA [x] -> phytanic acid [x] -> phytanic acid [c]. Notably, that loop hasn't existed since #648, but I've been testing/developing the tool on version 1.15 of Human-GEM to see if I can now automatically detect any of the problems I previously identified in a more ad-hoc fashion. While looking into that, I noticed all the problems in this issue, #774, #776, and #777

[Devlin-Moyer]

But also, after you suggested that I split #634 into multiple issues in #726, I noticed that several of the mitochondrial reactions that were associated with peroxisomal genes seemed like they should be removed for being duplicates of other reactions that didn't have peroxisomal genes associated with them, so the idea behind most of the recent issues about inconsistencies in fatty acid oxidation reactions is to get rid of all the mitochondrial reactions that shouldn't exist in the first place before worrying about removing the peroxisomal genes from their GPRs

[haowang-bioinfo]

I'm working on a new semi-automated model curation tool that identifies reactions that form loops capable of sustaining infinite fluxes, and one such loop was phytanic acid [c] -> phytanoyl-CoA [c] -> phytanoyl-CoA [x] -> phytanic acid [x] -> phytanic acid [c]. Notably, that loop hasn't existed since #648, but I've been testing/developing the tool on version 1.15 of Human-GEM to see if I can now automatically detect any of the problems I previously identified in a more ad-hoc fashion. While looking into that, I noticed all the problems in this issue, #774, #776, and #777

very good - could you please bring up this semi-automated tool as a discussion so that it could be eventually converted into GH action bot, which can avoid such cases from happening in the future while fixing the existing ones

[haowang-bioinfo]

the provided evidence seems fine and the proposed changes are rational, please move on to implementation

[Devlin-Moyer]

Fixed by #792