Duplicate Beta-Oxidation Reactions for (4Z,7Z,10Z,13Z,16Z)-docosapentaenoyl-CoA

[Devlin-Moyer]

Current behavior:

(4Z,7Z,10Z,13Z,16Z)-docosapentaenoyl-CoA can be beta-oxidized in a single step in
MAR03422: (4Z,7Z,10Z,13Z,16Z)-docosapentaenoyl-CoA [m] + 10 CoA [m] + 5 FAD [m] + 10 H2O [m] + 10 NAD+ [m] ⇒ 5 FADH2 [m] + 10 H+ [m] + 10 NADH [m] + 11 acetyl-CoA [m], GPR: ACAA1 or HSD17B10 or ACADVL or HADHA or ACADL or ACADM or ECHS1 or HADHB or HADH or ACAA2 or ACAD9

or a lot of steps:

ID	Reaction	Genes
MAR05380	(4Z,7Z,10Z,13Z,16Z)-docosapentaenoyl-CoA [m] + FAD [m] ⇒ 2,4,7,10,13,16-Docosahexenoyl Coenzyme A [m] + FADH2 [m]	ACADVL
MAR05404	2,4,7,10,13,16-Docosahexenoyl Coenzyme A [m] + H+ [m] + NADPH [m] ⇒ 3,7,10,13,16-Docosapentenoyl Coenzyme A [m] + NADP+ [m]	DECR1
MAR05383	3,7,10,13,16-Docosapentenoyl Coenzyme A [m] ⇒ 2,7,10,13,16-Docosapentenoyl Coenzyme A [m]	ECI1
MAR05378	2,7,10,13,16-Docosapentenoyl Coenzyme A [m] + CoA [m] + H2O [m] + NAD+ [m] ⇒ 5,8,11,14-Eicosatetraenoyl Coenzyme A [m] + H+ [m] + NADH [m] + acetyl-CoA [m]	HADHA and HADHB
MAR05373	5,8,11,14-Eicosatetraenoyl Coenzyme A [m] + CoA [m] + FAD [m] + H2O [m] + NAD+ [m] ⇒ 3,6,9,12-Octadecatetraenoyl Coenzyme A [m] + FADH2 [m] + H+ [m] + NADH [m] + acetyl-CoA [m]	ACADVL and HADHA and HADHB
MAR05367	3,6,9,12-Octadecatetraenoyl Coenzyme A [m] ⇒ 2,6,9,12-Octadecatetraenoyl Coenzyme A [m]	ECI1
MAR05360	2,6,9,12-Octadecatetraenoyl Coenzyme A [m] + CoA [m] + H2O [m] + NAD+ [m] ⇒ 4,7,10-Hexadecatrienoyl Coenzyme A [m] + H+ [m] + NADH [m] + acetyl-CoA [m]	HADHA and HADHB
MAR05195	4,7,10-Hexadecatrienoyl Coenzyme A [m] + FAD [m] ⇒ 2,4,7,10-Hexadecatetraenoyl Coenzyme A [m] + FADH2 [m]	ACADVL
MAR05302	2,4,7,10-Hexadecatetraenoyl Coenzyme A [m] + H+ [m] + NADPH [m] ⇒ 3,7,10-Hexadecatrienoyl Coenzyme A [m] + NADP+ [m]	DECR1
MAR05229	3,7,10-Hexadecatrienoyl Coenzyme A [m] ⇒ 2,7,10-Hexadecatrienoyl Coenzyme A [m]	ECI1
MAR05208	2,7,10-Hexadecatrienoyl Coenzyme A [m] + CoA [m] + H2O [m] + NAD+ [m] ⇒ 5,8-Tetradecadienoyl Coenzyme A [m] + H+ [m] + NADH [m] + acetyl-CoA [m]	HADHA and HADHB
MAR05123	5,8-Tetradecadienoyl Coenzyme A [m] + CoA [m] + FAD [m] + H2O [m] + NAD+ [m] ⇒ 3,6-Dodecadienoyl Coenzyme A [m] + FADH2 [m] + H+ [m] + NADH [m] + acetyl-CoA [m]	ACADVL and HADHA and HADHB
MAR05075	3,6-Dodecadienoyl Coenzyme A [m] ⇒ 2,6-Dodecadienoyl Coenzyme A [m]	ECI1
MAR05072	2,6-Dodecadienoyl Coenzyme A [m] + CoA [m] + H2O [m] + NAD+ [m] ⇒ 4-cis-decenoyl-CoA [m] + H+ [m] + NADH [m] + acetyl-CoA [m]	ECHS1 and HADH and ACAA2
MAR03294	4-cis-decenoyl-CoA [m] + FAD [m] ⇒ 2-trans-4-cis-decadienoyl-CoA [m] + FADH2 [m]	ACADVL or ACADL or ACADM or ACAD9
MAR04970	2-trans-4-cis-decadienoyl-CoA [m] + H+ [m] + NADPH [m] ⇒ NADP+ [m] + trans-3-decenoyl-CoA [m]	DECR1
MAR03298	(2E)-decenoyl-CoA [m] ⇔ trans-3-decenoyl-CoA [m]	ECI1 or ECI2
MAR03143	(2E)-decenoyl-CoA [m] + H2O [m] ⇒ (S)-hydroxydecanoyl-CoA [m]	HADHA and EHHADH and ECHS1
MAR03144	(S)-hydroxydecanoyl-CoA [m] + NAD+ [m] ⇒ 3-oxodecanoyl-CoA [m] + H+ [m] + NADH [m]	HSD17B10 and HADHA and HADH
MAR03146	3-oxodecanoyl-CoA [m] + CoA [m] ⇒ acetyl-CoA [m] + octanoyl-CoA [m]	HADHA and HADHB and ACAA2
MAR03149	FAD [m] + octanoyl-CoA [m] ⇒ (2E)-octenoyl-CoA [m] + FADH2 [m]	ACADM or ACAD9
MAR03150	(2E)-octenoyl-CoA [m] + H2O [m] ⇒ (S)-hydroxyoctanoyl-CoA [m]	HADHA and EHHADH and ECHS1
MAR03151	(S)-hydroxyoctanoyl-CoA [m] + NAD+ [m] ⇒ 3-oxooctanoyl-CoA [m] + H+ [m] + NADH [m]	HSD17B10 and HADHA and HADH
MAR03153	3-oxooctanoyl-CoA [m] + CoA [m] ⇒ acetyl-CoA [m] + hexanoyl-CoA [m]	HADHA and HADHB and ACAA2
MAR03156	FAD [m] + hexanoyl-CoA [m] ⇒ (2E)-hexenoyl-CoA [m] + FADH2 [m]	ACADSB
MAR03157	(2E)-hexenoyl-CoA [m] + H2O [m] ⇒ (S)-hydroxyhexanoyl-CoA [m]	HADHA and EHHADH and ECHS1
MAR03158	(S)-hydroxyhexanoyl-CoA [m] + NAD+ [m] ⇒ 3-oxohexanoyl-CoA [m] + H+ [m] + NADH [m]	HSD17B10 and HADHA and HADH
MAR03160	3-oxohexanoyl-CoA [m] + CoA [m] ⇒ acetyl-CoA [m] + butyryl-CoA [m]	HADHA and HADHB and ACAA2
MAR03163	FAD [m] + butyryl-CoA [m] ⇒ FADH2 [m] + crotonyl-CoA [m]	ACADM or ACADS or ACADSB or ACAD8 or ACAD9 or ACAD11
MAR03164	H2O [m] + crotonyl-CoA [m] ⇒ (S)-3-hydroxybutyryl-CoA [m]	HADHA and EHHADH and ECHS1
MAR03166	(S)-3-hydroxybutyryl-CoA [m] + NAD+ [m] ⇒ H+ [m] + NADH [m] + acetoacetyl-CoA [m]	HSD17B10 and HADHA and HADH
MAR03885	CoA [m] + acetoacetyl-CoA [m] ⇔ 2 acetyl-CoA [m]	ACAT1

Expected feature/value/output:

Since MAR05380 - MAR03885 accurately represent each individual step of the beta-oxidation of (4Z,7Z,10Z,13Z,16Z)-docosapentaenoyl-CoA in much more detail than MAR03422, I think MAR03422 should be removed for being unnecessary.

[JonathanRob]

Really enjoying this clean-up of the beta-oxidation reactions. This looks good to me.

Out of curiosity, and maybe I missed it before, are you using a script-based approach to search for these cases where reactions are the same but broken into multiple steps, or are you manually identifying them as you go?

[Devlin-Moyer]

I used a script to identify all reactions that had metabolites in [m] that were associated with one or more genes that encode exlusively peroxisomal beta-oxidation enzymes in #634, but eventually realized that many of those reactions had other problems, so all of these issues are manual follow ups. For this issue in particular, I identified a whole set (maybe like 15-20) of reactions like MAR03422 that do complete beta-oxidation of something in a single step, but so far, each individual case has a somewhat unique set of problems associated with it, so there are going to be a lot of issues following up on those. (beta-oxidation of linoleoyl- and gamma-linolenoyl-CoA is particularly messy)

[JonathanRob]

Ok, great work though.

[haowang-bioinfo]

I used a script to identify all reactions that had metabolites in [m] that were associated with one or more genes that encode exlusively peroxisomal beta-oxidation enzymes in #634, but eventually realized that many of those reactions had other problems, so all of these issues are manual follow ups. For this issue in particular, I identified a whole set (maybe like 15-20) of reactions like MAR03422 that do complete beta-oxidation of something in a single step, but so far, each individual case has a somewhat unique set of problems associated with it, so there are going to be a lot of issues following up on those. (beta-oxidation of linoleoyl- and gamma-linolenoyl-CoA is particularly messy)

good to know this - would be nice to include the script that can potentially be turned into a GH action bot at some point

[Devlin-Moyer]

ah yea I just did something like

perox_genes = ['ACOX1', 'ACOX2', 'ACOX3', 'ACAD11', 'EHHADH', 'HSD17B4', 'ACAA1', 'DECR2']
for r in model.reactions:
    if any(g in r.gene_name_reaction_rule for g in perox_genes) and all(m.compartment == 'm' for m in r.metabolites):
        print(f'{r.id} | {r.build_reaction_string(True)} | {r.gene_name_reaction_rule}')

If you wanted to scale that up to a GH action, I suppose you could take the localization info in model/genes.tsv and verify that each gene was only associated with reactions in compartments it was known to be localized to. However, I'm pretty sure it would take a significant amount of curation to get the existing model to pass that test; I'm currently working on an issue to address the fact that ECI1 and HADHA (both exclusively mitochondrial beta-oxidation enzymes) are associated with several dozen peroxisomal reactions, and I gotta y'know see if they're already associated with the correct peroxisomal genes or if removing the clearly incorrect mitochondrial genes would leave them associated with no genes.

[haowang-bioinfo]

If you wanted to scale that up to a GH action, I suppose you could take the localization info in model/genes.tsv and verify that each gene was only associated with reactions in compartments it was known to be localized to.

scaling this up to GH action would be a very good idea that worth further consideration