Suspected duplicated metabolites from MetaNetX #626

Activity

1. JHL-452b

   added

   bugThe model or code is wrongThe model or code is wrong

   on Jun 9, 2023
2. feiranl commented on Jun 20, 2023

   

   feiranl

   on Jun 20, 2023

   Last edited by feiranl

   Collaborator

   More actions

   @haowang-bioinfo Is that possible that we can turn to MNX group for help on globally updating MNX IDs?
3. haowang-bioinfo commented on Aug 13, 2023

   

   haowang-bioinfo

   on Aug 13, 2023

   Contributor

   More actions

   @haowang-bioinfo Is that possible that we can turn to MNX group for help on globally updating MNX IDs?

   @feiranl good idea - @smoretti, @ANiknejad if you have any code for consistency check?
4. edkerk

   added 3 commits that reference this issue on Jul 13, 2026

   chore: annotation overhaul plan and structure/formula/charge audit

   ...

   7830601

   chore: annotation overhaul plan and structure/formula/charge audit

   ...

   530806a

   feat: structure-verified annotation overhaul (metabolite + reaction c…

   ...

   43308d7
5. edkerk

   added

   wipA pull request for this issue is openA pull request for this issue is open

   on Sep 24, 2026
6. edkerk commented on Sep 26, 2026

   

   edkerk

   on Sep 26, 2026

   Last edited by edkerk

   Member

   More actions

   Update, checked against #1090. Metabolite IDs are grouped by identical structure (InChIKey including stereochemistry, ignoring protonation). 74 groups remain.

   Same formula and charge (52), mostly true duplicates to merge. For example:

   • MAM00623 / MAM00625 (26- and 27-hydroxycholesterol, the same compound)
   • MAM00106 / MAM02391 (linoleoyl-CoA)
   • MAM01364 / MAM03571 (arachidonyl-CoA)
   • MAM00008 / MAM03569 (eicosadienoic acid)
   • MAM00103 / MAM03284, MAM00074 / MAM03171, MAM00118 / MAM03972, MAM03203 / MAM03967, MAM01576 / MAM03978 / MAM04041, MAM00678 / MAM03543, MAM03552 / MAM03553 (acyl-CoAs)
   • MAM00653 / MAM03188 (2-hydroxyglutarate)
   • MAM01016 / MAM03276 (4-methylthio-2-oxobutanoate)
   • MAM00392 / MAM01892 (15-keto-PGF2α)
   • MAM00993 / MAM01006 (4-hydroxyretinoate)
   • MAM01067 / MAM04082 (zymosterol intermediate)
   • MAM00942 / MAM03331
   • MAM02387 / MAM03787 (linoleate)
   • MAM02640 / MAM03790
   • MAM02723 / MAM03862 (phenylacetylglycine)
   • MAM03685 / MAM10030 (hyocholate)
   • MAM03996 / MAM10026 (taurohyocholate)
   • MAM01392 / MAM03227 (3-hydroxyisovalerate)
   • MAM01468 / MAM04079
   • MAM03231 / MAM03232 (merged in open PR fix: QC check for reactions split across compartments, and curation of those found #1095)
   • Pairs added twice recently: MAM20034 / MAM20035, MAM20041 / MAM20042, MAM20049 / MAM20050, MAM20051 / MAM20052, MAM20054 / MAM20055, MAM20056 / MAM20057.

   Some groups in this set differ only in stereochemistry that their structures do not record, so they are not duplicates; their structure needs fixing (#1085):

   • MAM02747 / MAM03362 ((3R)/(3S)-phytanoyl-CoA)
   • MAM01201 / MAM01206 (8R/8S)
   • MAM10035 / MAM10036 / MAM10037 (α/β/ω-muricholoyl-CoA)
   • MAM00750 / MAM00751
   • MAM01090 / MAM01092
   • MAM00252 / MAM01150

   Different formula or charge (22). Some are acid/base pairs, which is correct: NH3 / NH4+, H2O / OH-, H2S2O3 / thiosulfate. Most point to a wrong structure on one side (#1085), e.g.:

   • MAM03621 / MAM03714 (glycyl-glycine / leucyl-glycine)
   • MAM02153 / MAM02587 (hydroxy-nifedipine / nifedipine)
   • MAM02941 / MAM03792 (stearoyl- / octadecenoyl-CoA)
   • MAM00292 / MAM00303 (11-cis-retinyl palmitate / 11-peroxy-DHA)
   • MAM01450 / MAM01511 (cholesterol / cholesterol-STAR)
   • MAM02641 / MAM02642 (octanoate radical / octanoic acid)
7. edkerk commented on Sep 26, 2026

   

   edkerk

   on Sep 26, 2026

   Member

   More actions

   The same-formula groups have been reviewed together with their reactions. #1090 now merges 37 of them. Some groups are not merged:

   • stereoisomers or different compounds whose structures in the model are identical, so the structure needs fixing (Curate remaining metabolite structure, formula and identity inconsistencies #1085): (6Z,9Z)-octadecadienoyl-CoA vs linoleoyl-CoA; 10,11- vs 6,7-dihydro-12-epi-LTB4; 12(S)-HETE vs CYP2F1's 12-hydroxy-arachidonate; the 13-cis/9-cis/generic retinoyl glucuronides; 8R/8S; elaidate/oleate; 3R/3S-phytanoyl-CoA; selenohomocysteine vs Se-methylselenocysteine; the geranylgeranyl-PP isomers; α/β/ω-muricholoyl-CoA; (2S)- vs generic pristanoyl-CoA;
   • the cis and trans 5,8-tetradecadienoyl-CoAs and 3,6-dodecadienoyl-CoAs, which belong to different β-oxidation routes;
   • MAM03231/MAM03232, merged in fix: QC check for reactions split across compartments, and curation of those found #1095, and MAM20018/20021 and MAM20026/20027, deprecated in fix: use NADPH in CYP2E1 fatty-acid hydroxylations #1094.
8. edkerk

   added a commit that references this issue on Sep 26, 2026

   fix: merge MAM20017n into MAM20015 (13,14-dihydro-15-oxo-PGE2)

   ...

   fb78e7c
9. edkerk commented on Sep 27, 2026

   

   edkerk

   on Sep 27, 2026

   Member

   More actions

   Follow-up to #1090: #1105 corrects several side effects of the merges, and #1104 merges one pair that was missed.

   GPR unions. Merging identical reactions joined their GPRs with OR. For some reactions this added genes with no evidence for that step; they are removed again.

   Reaction	GPR after #1090	GPR now	Evidence
   MAR01762 cholesterol 26-hydroxylation [m]	CYP27A1 or CH25H	CYP27A1	CH25H only 25-hydroxylates (UniProt O95992, PMID:9852097); CYP27A1 is the sterol 26(27)-hydroxylase (UniProt Q02318, PMID:1708392)
   MAR01765 26-hydroxycholesterol 7alpha-hydroxylation	CYP7B1 or CYP2E1 or CYP27A1	CYP7B1	7alpha-hydroxylation of 26-hydroxycholesterol is CYP7B1 (PMID:10588945); CYP2E1 and CYP27A1 do not 7alpha-hydroxylate
   MAR01605 5beta-cholestane-3alpha,7alpha,12alpha-triol 26-hydroxylation [m]	CYP27A1 or CYP2E1 or CYP27A1	CYP27A1	CYP27A1 as above; no CYP2E1 activity on this bile-acid intermediate
   MAR06700 all-trans-retinoate 4-hydroxylation	CYP26A1 or CYP2J2 or CYP2C9 or CYP2C8	CYP26A1 or CYP2C9 or CYP2C8	CYP2J2 has no retinoid activity in UniProt (P51589); the only CYP2J data are for sheep (PMID:20021200). Kept: CYP26A1 (PMID:22020119), CYP2C8 and CYP2C9 (human liver microsomes and expressed enzymes, PMID:10484078, PMID:11093772)
   MAR01611 trihydroxycholestanal -> cholestanoate	CYP27A1 or ALDH3A1 or ALDH9A1	CYP27A1	no report of ALDH3A1/ALDH9A1 with this bile-acid aldehyde; CYP27A1 catalyses it (RHEA:34627, PMID:1708392, PMID:11412116)
   MAR04049 DMBA-3,4-diol epoxidation	10 P450s	CYP1B1 or CYP1A1	only CYP1A1/CYP1B1 established for this step (PMID:8555413, PMID:8674051)

   Genes listed twice in joined GPRs are removed on MAR01540, MAR01605, MAR01805, MAR10101, MAR10103 and MAR10119 (no change in meaning). #1106 also deprecates MAR01785 and MAR01781, where CH25H was given 26- and 20-hydroxylation.

   Annotations of kept entities

   • MAR08607 (glutamine + 4-methylthio-2-oxobutanoate -> 2-oxoglutaramate + methionine, KYAT1) kept the ids of its old equation (BiGG/Recon3D UNK2, MNXR105128, rxn10944, RCR10315). It now has RHEA:30392 (master RHEA:30391), MNXR119519, KEGG R12507, rxn41389, and EC 2.6.1.117;2.6.1.64. Glutamine transaminase K closes methionine salvage (rat, PMID:7068632; review PMID:15016471; human KYAT1 substrate range, PMID:15606768).
   • HCl (MAM02144c): MNXM56769 is the "hydrochloride" salt class without a formula; MetaNetX maps HCl's ChEBI and KEGG ids to MNXM735978, which is restored. fix: pre-existing reaction and GPR errors found in the curation review #1106 then replaces HCl by chloride + H+ in its reactions and deprecates MAM02144 into chloride.
   • MAM03543 was renamed 2-trans-4-cis-decadienoyl-CoA (made from 4-cis-decenoyl-CoA and reduced by DECR1; PMID:15531764, PMID:2332510) but kept a (2E,4E) structure and ids. It now has the (2E,4Z) structure, CHEBI:137593, MNXM1601, PubChem 6444117 (c, m, x). Its precursor MAM00980 had the same (4E) problem; that is fixed in fix: pre-existing metabolite identity, structure and cross-reference errors #1107.
   • metRetired: the 14 renamed ids now list the id they replace (MAM03971e on MAM03970e, and MAM20005c->MAM20003c, MAM20006c->MAM20004c, MAM20009c->MAM20007c, MAM20010c->MAM20008c, MAM20016n->MAM20014n, MAM20025i->MAM20024i, MAM20029i->MAM20028i, MAM20031i->MAM20030i, MAM20033c->MAM20032c, MAM20044c->MAM20043c, MAM20046c->MAM20045c, MAM20048c->MAM20047c, MAM20053r->MAM20013r), as earlier renames do.
   • Names: where the kept id had the less precise name of a merged pair, it gets the better one (14 metabolites), e.g. MAM04082 "Zymosterol Intermediate 2" -> 5alpha-cholesta-8,24-dien-3-one, MAM00384 "15-deoxy-PGD2" -> 15-deoxy-delta-12,14-PGD2, MAM00751 "...-27-al" -> (25R)-3alpha,7alpha,12alpha-trihydroxy-5beta-cholestan-26-al, MAM02640 "octadecenoylcarnitine(7)" -> (11Z)-octadecenoylcarnitine, MAM01364 -> arachidonoyl-CoA. Each name matches the model structure and a ChEBI entry.

   Reactions

   • MAR00751 (ACADM, 4,8-dimethylnonanoyl-CoA oxidation with 2 H+ in place of FAD) became a charge-imbalanced near-duplicate of MAR06346. It is deprecated into MAR06346, which gets ACADM and records MAR00751 in rxnRetired.
   • MAR00036 ("(11R)-HPETE to arachidonate derivatives conversion", a pool reaction) consumed 2 of MAM00384c after the 15-deoxy-PGD2 merge; now 1.

   Missed pair (in #1104). MAM20015g and MAM20017n are both 13,14-dihydro-15-oxo-PGE2 (MNXM733372). MAM20017n is renamed MAM20015n, with charge -1 as for MAM20015g, and added to deprecatedMetabolites.tsv.

   Structures that were identical for different compounds (listed in the earlier comment as not merged): (6Z,9Z)-octadecadienoyl-CoA, 10,11-dihydro-12-epi-LTB4, alpha-/omega-muricholoyl-CoA, 4-cis-decenoyl-CoA and selenohomocysteine now have their own structures and ids (#1105 and #1107; details in #1115).
10. edkerk

    added a commit that references this issue on Sep 28, 2026

    fix: curate metabolite structure, formula and identity inconsistencies (

    ...

    ca86002
11. edkerk

    added

    fixed in developMerged into develop, will be in the next releaseMerged into develop, will be in the next release

    and removed

    wipA pull request for this issue is openA pull request for this issue is open

    on Sep 28, 2026
12. smoretti commented on Sep 28, 2026

    

    smoretti

    on Sep 28, 2026

    More actions

    @feiranl good idea - @smoretti, @ANiknejad if you have any code for consistency check?

    @feiranl @haowang-bioinfo Sorry for the very late reply. The consistency checks are done internally in our pipeline during the reconciliation, not really available outside.

    We are currently integrating MetabolicAtlas metabolites and reactions, using Human-GEM, in MetaNetX.
    Maybe we should wait for the duplication cleaning to be completed.
13. edkerk

    added this to the 2.1.0 milestone on Sep 28, 2026
14. edkerk commented on Oct 5, 2026

    

    edkerk

    on Oct 5, 2026

    Member

    More actions

    Fixed in Human 2.1.0.
15. edkerk

    closed this as completedon Oct 5, 2026
16. edkerk

    removed

    fixed in developMerged into develop, will be in the next releaseMerged into develop, will be in the next release

    on Oct 14, 2026