source:
  candidates: projects/RAR_MISANNOTATION/data/rar_rerun_direct_candidates.yaml
  triage_date: '2026-11-11'
  status: FIRST_PASS
  note: >-
    First-pass verdicts from the protein's curated UniProt function text, the GO term
    definition and its taxon constraints, and family knowledge. The cited papers were not
    read for this pass, so no verdict here is a gene review, and nothing is filed upstream.
    Verdicts: ANNOTATION_WRONG (the term does not apply to this protein), CONSTRAINT_QUESTIONABLE
    (the annotation looks right and the GO taxon constraint looks too narrow), FLAG_FAILS
    (the low rate is a curation-density artefact; the annotation is right), TRUE_BUT_GENERAL
    (right, but a more specific term fits; a MODIFY candidate), UNCERTAIN. Where a verdict
    rests on a cached paper or UniProt record, supported_by gives the verbatim text.
# Tranche A: candidate annotations that break a taxon constraint GO already has. These are
# mechanical; the only question is whether the annotation or the constraint is wrong.
constraint_violations:
- go_id: GO:0004407
  label: histone deacetylase activity
  constraint: never_in_taxon NCBITaxon:2 (Bacteria)
  annotations: [UniProtKB:P39067 AcuC, B. subtilis, IBA]
  verdict: ANNOTATION_WRONG
  reason: Bacteria have no histones. AcuC deacetylates Lys of acetyl-CoA synthetase AcsA (UniProt
    FUNCTION, PubMed:16855235), so the right activity is GO:0033558 protein lysine deacetylase
    activity. The IBA crosses a constraint GO already has.
  supported_by:
  - reference_id: PMID:16855235
    supporting_text: Unlike sirtuins, the AcuC protein did not require NAD(+) as cosubstrate to
      deacetylate AcsA(Ac).
- go_id: GO:0000290
  label: deadenylation-dependent decapping of nuclear-transcribed mRNA
  constraint: only_in_taxon NCBITaxon:2759 (Eukaryota)
  annotations: [UniProtKB:P0A776 RppH E. coli, UniProtKB:O25826 H. pylori, UniProtKB:P65554
      S. typhimurium; all IBA from PANTHER:PTN000561806]
  verdict: ANNOTATION_WRONG
  reason: Bacterial RppH removes 5' pyrophosphate from triphosphorylated RNA to start 5'-end-dependent
    decay (UniProt FUNCTION, PubMed:18202662); bacteria have neither nuclear transcription nor
    deadenylation-dependent decapping. The PTN000561806 node is placed above the bacterial clade.
  supported_by:
  - reference_id: PMID:18202662
    supporting_text: In vitro, RppH efficiently removes pyrophosphate from the 5' end of triphosphorylated
      RNA
- go_id: GO:0005813
  label: centrosome
  constraint: never_in_taxon NCBITaxon:33090 (Viridiplantae)
  annotations: [UniProtKB:Q9M2Q7 Arabidopsis, UniProtKB:D8RR71 Selaginella; IBA from PANTHER:PTN000471210]
  verdict: ANNOTATION_WRONG
  reason: Land plants have no centrosomes; the IBA node sits above the plant branch.
- go_id: GO:0043066
  label: negative regulation of apoptotic process
  constraint: only_in_taxon NCBITaxon:33154 (Opisthokonta)
  annotations: [UniProtKB:Q9C901 and UniProtKB:D8SSK8 (IBA from PANTHER:PTN001580765), UniProtKB:Q9SNC6
      (IMP, CACAO)]
  verdict: ANNOTATION_WRONG
  reason: GO keeps apoptosis to opisthokonts; plant cell death is annotated to GO:0043069 negative
    regulation of programmed cell death.
- go_id: GO:0043065
  label: positive regulation of apoptotic process
  constraint: only_in_taxon NCBITaxon:33154 (Opisthokonta)
  annotations: [UniProtKB:Q9LSF1 Arabidopsis (IMP, TAIR)]
  verdict: ANNOTATION_WRONG
  reason: As for GO:0043066; the plant term is GO:0043068 positive regulation of programmed cell death.
- go_id: GO:0043682
  label: P-type divalent copper transporter activity
  constraint: only_in_taxon NCBITaxon:2 (Bacteria)
  annotations: [UniProtKB:Q04656 ATP7A human (ISS), UniProtKB:Q64446 and UniProtKB:B1AQ57 mouse (ISO)]
  verdict: ANNOTATION_WRONG
  reason: ATP7A and ATP7B pump Cu(I); the matching term is GO:0140581 P-type monovalent copper transporter
    activity. The ISS/ISO rows inherit the wrong valence.
- go_id: GO:0071266
  label: "'de novo' L-methionine biosynthetic process"
  constraint: never_in_taxon NCBITaxon:40674 (Mammalia)
  annotations: [UniProtKB:Q93088 BHMT human (IMP), UniProtKB:O35490 mouse (ISO)]
  verdict: ANNOTATION_WRONG
  reason: BHMT remethylates homocysteine, which regenerates methionine; mammals cannot make it de novo.
    GO:0071267 L-methionine salvage ("generates L-methionine from derivatives of it, without de novo
    synthesis") fits exactly.
- go_id: GO:0030682
  label: symbiont-mediated perturbation of host defenses
  constraint: never_in_taxon NCBITaxon:7742 (Vertebrata)
  annotations: [UniProtKB:A2ADZ6, UniProtKB:F8WHL4, UniProtKB:O35345 mouse (IMP, MGI, with MGI:MGI:3856636)]
  verdict: ANNOTATION_WRONG
  reason: A mouse gene product cannot be the symbiont; the host side has its own defense-response terms.
- go_id: GO:0020002
  label: host cell plasma membrane
  constraint: never_in_taxon NCBITaxon:7742 (Vertebrata)
  annotations: [UniProtKB:Q5T0T0 MARCHF8 human (NAS, UniProt)]
  verdict: ANNOTATION_WRONG
  reason: A human protein's membrane is not a host cell membrane; plasma membrane is the term.
- go_id: GO:0044732
  label: mitotic spindle pole body
  constraint: only_in_taxon NCBITaxon:4751 (Fungi)
  annotations: [UniProtKB:P07911 UMOD human (IDA, PMID:20172860, assigned by MGI)]
  verdict: ANNOTATION_WRONG
  reason: The spindle pole body is the fungal MTOC, so the term cannot apply to a human protein. The
    paper does report UMOD at mitotic spindle poles in human cells, so the row is a wrong-term choice,
    not a mis-keyed one; the animal counterpart is GO:0097431 mitotic spindle pole. A spindle-pole
    location for a secreted GPI-anchored protein is unusual, so the full text (not cached) should be
    read before the claim is used for more than this term swap.
  supported_by:
  - reference_id: PMID:20172860
    supporting_text: UMOD localization at the mitotic spindle poles and colocalization with other ciliary
      proteins such as nephrocystin-1 and kinesin family member 3A is demonstrated.
- go_id: GO:0010973
  label: positive regulation of division septum assembly
  constraint: only_in_taxon NCBITaxon:2 or NCBITaxon:4751 (Fungi or Bacteria)
  annotations: [UniProtKB:Q9N4A7 npp-20 worm (IMP, UniProt)]
  verdict: ANNOTATION_WRONG
  reason: Animal cells form no division septum. The worm npp-20 review already MODIFYs this row
    (genes/worm/npp-20/npp-20-ai-review.yaml), from the first RAR round.
- go_id: GO:0016477
  label: cell migration
  constraint: never_in_taxon NCBITaxon:4890 (Ascomycota)
  annotations: [UniProtKB:P13649 URA3, UniProtKB:Q59X67; Candida albicans (IMP, CGD)]
  verdict: ANNOTATION_WRONG
  reason: Walled fungal cells do not migrate; invasive or chemotropic hyphal growth has its own terms.
    URA3 is orotidine 5'-phosphate decarboxylase, so a cell-migration phenotype is indirect at best.
- go_id: GO:0050728
  label: negative regulation of inflammatory response
  level: phylum
  constraint: only_in_taxon NCBITaxon:7742 (Vertebrata)
  annotations: [UniProtKB:P34021 fly, UniProtKB:G5ECR9 worm; IBA]
  verdict: ANNOTATION_WRONG
  reason: Inflammation is a vertebrate process in GO; the IBA node is placed above the invertebrate branches.
- go_id: GO:0061028
  label: establishment of endothelial barrier
  level: phylum
  constraint: only_in_taxon NCBITaxon:7742 (Vertebrata)
  annotations: [UniProtKB:O46101 fly (IMP, UniProt)]
  verdict: ANNOTATION_WRONG
  reason: Insects have no endothelium; an epithelial barrier term would fit.
- go_id: GO:0022849
  label: glutamate-gated calcium ion channel activity
  constraint: only_in_taxon NCBITaxon:33208 (Metazoa)
  annotations: [UniProtKB:Q9C8E7 GLR3.3 Arabidopsis (IMP, TAIR)]
  verdict: CONSTRAINT_QUESTIONABLE
  reason: Plant GLRs are glutamate-receptor-like, ligand-gated cation channels, and GLR3.3 conducts
    calcium in response to glutamate. A Metazoa-only constraint looks too narrow; check the TAIR paper
    before proposing a change.
- go_id: GO:0035235
  label: ionotropic glutamate receptor signaling pathway
  constraint: only_in_taxon NCBITaxon:33208 (Metazoa)
  annotations: [UniProtKB:Q9C8E7 GLR3.3 Arabidopsis (IEA, GOC, inferred from GO:0022849)]
  verdict: CONSTRAINT_QUESTIONABLE
  reason: Follows from the GO:0022849 row by GO's inference rule, so it stands or falls with that
    constraint.
- go_id: GO:0001778
  label: plasma membrane repair
  constraint: only_in_taxon NCBITaxon:33208 (Metazoa)
  annotations: [UniProtKB:Q9SKR2 SYT1 Arabidopsis (IMP, TAIR)]
  verdict: CONSTRAINT_QUESTIONABLE
  reason: Arabidopsis synaptotagmin 1 is studied as a plasma membrane integrity and resealing factor
    under freezing and mechanical stress, a plant counterpart of animal membrane repair.
- go_id: GO:0015485
  label: cholesterol binding
  constraint: never_in_taxon NCBITaxon:4751 (Fungi)
  annotations: [UniProtKB:Q59ZX3 RBE1 (PRY family), UniProtKB:Q5AB48; Candida albicans (IDA, CGD)]
  verdict: CONSTRAINT_QUESTIONABLE
  reason: CAP/PRY-family proteins bind sterols, cholesterol included, in vitro; a binding activity
    can be measured whether or not the organism makes cholesterol.
- go_id: GO:0007288
  label: sperm axoneme assembly
  constraint: only_in_taxon NCBITaxon:33208 (Metazoa)
  annotations: [UniProtKB:D8STG0, UniProtKB:D8T5A3 Selaginella (IBA)]
  verdict: CONSTRAINT_QUESTIONABLE
  reason: Lycophytes such as Selaginella make flagellate sperm, so a Metazoa-only constraint excludes
    real cases. Whether these particular IBAs hold needs the PAINT node.
- go_id: GO:0042246
  label: tissue regeneration
  constraint: only_in_taxon NCBITaxon:33208 (Metazoa)
  annotations: [UniProtKB:Q8W485 (IMP), UniProtKB:Q8L5Z1 (IEP); Arabidopsis, TAIR]
  verdict: UNCERTAIN
  reason: Plants regenerate tissues, but GO may intend plant regeneration to go under plant-specific
    terms; either the annotation or the constraint needs to move.
- go_id: GO:0003151
  label: outflow tract morphogenesis
  level: phylum
  constraint: only_in_taxon NCBITaxon:7742 (Vertebrata)
  annotations: [UniProtKB:P24014, UniProtKB:Q24298 (IMP), UniProtKB:M9PC46 (IGI); fly, FlyBase]
  verdict: UNCERTAIN
  reason: The Drosophila heart has a described outflow region; whether it is homologous enough for this
    term is a question for FlyBase and GO.
# Tranche B: IBA flags in prokaryotes (the paper's Udg pattern), by PAINT node.
prokaryote_iba:
- node: PANTHER:PTN007723485
  terms: [GO:0043161 proteasome-mediated ubiquitin-dependent protein catabolic process, GO:1990756
      ubiquitin-like ligase-substrate adaptor activity]
  annotations: [UniProtKB:O31912 YorA, UniProtKB:P94409 YclG; B. subtilis]
  verdict: ANNOTATION_WRONG
  reason: B. subtilis has neither ubiquitin nor a 26S proteasome; both proteins are uncharacterized.
- node: PANTHER:PTN000561806
  terms: [GO:0000290 deadenylation-dependent decapping of nuclear-transcribed mRNA]
  annotations: [RppH in E. coli, H. pylori, S. typhimurium]
  verdict: ANNOTATION_WRONG
  reason: Also a constraint violation; see tranche A.
- node: PANTHER:PTN008143312
  terms: [GO:0004407 histone deacetylase activity]
  annotations: [UniProtKB:P39067 AcuC, B. subtilis]
  verdict: ANNOTATION_WRONG
  reason: Also a constraint violation; see tranche A.
- node: PANTHER:PTN000155236
  terms: [GO:0017101 aminoacyl-tRNA synthetase multienzyme complex]
  annotations: [UniProtKB:P47525 ProS, M. genitalium]
  verdict: ANNOTATION_WRONG
  reason: The multi-synthetase complex is eukaryotic; nothing places Mycoplasma ProS in one.
- node: PANTHER:PTN005367244
  terms: [GO:0016125 sterol metabolic process]
  annotations: [UniProtKB:O34539 YjiC B. subtilis, UniProtKB:Q8ZMN4 S. typhimurium]
  verdict: ANNOTATION_WRONG
  reason: Neither organism makes sterols. YjiC is a promiscuous glycosyltransferase of flavonoids and
    other aromatics (UniProt FUNCTION, PubMed:28315700).
  supported_by:
  - reference_id: PMID:28315700
    supporting_text: Bs-YjiC, with uridine diphosphate glucose (UDPG) as sugar donor, exhibited robust
      capabilities to glycosylate 19 structurally diverse types of drug-like scaffolds
- node: PANTHER:PTN000770577
  terms: [GO:0016117 carotenoid biosynthetic process]
  annotations: [UniProtKB:O06728 YisP, B. subtilis]
  verdict: ANNOTATION_WRONG
  reason: YisP is a farnesyl diphosphate phosphatase that makes farnesol for biofilm formation and was
    first mistaken for a squalene synthase (UniProt FUNCTION, PubMed:25308276).
  supported_by:
  - reference_id: PMID:25308276
    supporting_text: We find that YisP acts as a phosphatase, catalyzing formation of farnesol from farnesyl
      diphosphate
- node: PANTHER:PTN002682344
  terms: [GO:0140378 protein complex scaffold activity]
  annotations: [UniProtKB:P62672 ApaG E. coli, UniProtKB:Q56017 S. typhimurium]
  verdict: UNCERTAIN
  reason: Bacterial ApaG has no characterized function; the scaffold activity comes from the eukaryotic
    ApaG-domain proteins, so the transfer is probably over-propagated.
- node: PANTHER:PTN002203706
  terms: [GO:0050545 sulfopyruvate decarboxylase activity]
  annotations: [UniProtKB:O06738 ComB, B. subtilis]
  verdict: UNCERTAIN
  reason: UniProt names ComB a probable 2-phosphosulfolactate phosphatase, a different step of coenzyme M
    synthesis; the decarboxylase is ComDE in archaea. Needs the PANTHER family checked.
- node: PANTHER:PTN000222418
  terms: [GO:0004421 hydroxymethylglutaryl-CoA synthase activity]
  annotations: [UniProtKB:P40830 PksG, B. subtilis]
  verdict: UNCERTAIN
  reason: PksG makes an HMG unit on an acyl carrier protein during bacillaene synthesis (UniProt FUNCTION,
    PubMed:16757561); the CoA-substrate term is a near miss.
  supported_by:
  - reference_id: PMID:16757561
    supporting_text: they act to incorporate an acetate-derived beta-methyl branch on an acetoacetyl-S-carrier
      protein
- node: PANTHER:PTN008451158
  terms: [GO:0004737 pyruvate decarboxylase activity]
  annotations: [UniProtKB:H9L492, S. typhimurium]
  verdict: UNCERTAIN
  reason: An uncharacterized thiamine diphosphate enzyme; Salmonella is not known for pyruvate decarboxylase.
- node: PANTHER:PTN008551622
  terms: [GO:0072659 protein localization to plasma membrane]
  annotations: [UniProtKB:O32076 FloT, B. subtilis]
  verdict: UNCERTAIN
  reason: Bacterial flotillins organize membrane microdomains (UniProt FUNCTION); a role in localizing
    their client proteins is plausible but not shown here.
- node: PANTHER:PTN002185576
  terms: [GO:0004867 serine-type endopeptidase inhibitor activity]
  annotations: [UniProtKB:P23827 ecotin E. coli, UniProtKB:Q8ZNH4 S. typhimurium]
  verdict: FLAG_FAILS
  reason: Ecotin inhibits chymotrypsin, trypsin, elastases and other serine proteases (UniProt FUNCTION,
    PubMed:35483450). Rare in bacteria, but correct.
  supported_by:
  - reference_id: PMID:35483450
    supporting_text: Escherichia coli ecotin inhibits trypsin-like, chymotrypsin-like, and elastase-like
      enzymes
- node: PANTHER:PTN000644795
  terms: [GO:0047475 phenylacetate-CoA ligase activity]
  annotations: [UniProtKB:P76085 PaaK, E. coli]
  verdict: FLAG_FAILS
  reason: PaaK activates phenylacetate to phenylacetyl-CoA (PubMed:20660314); see genes/ECOLI/paaK.
  supported_by:
  - reference_id: file:ECOLI/paaK/paaK-uniprot.txt
    supporting_text: Catalyzes the activation of phenylacetic acid (PA) to phenylacetyl-CoA (PA-CoA).
- node: PANTHER:PTN001627042
  terms: [GO:0006303 double-strand break repair via nonhomologous end joining]
  annotations: [UniProtKB:O34398 LigD, B. subtilis]
  verdict: FLAG_FAILS
  reason: Bacteria have a Ku-LigD NHEJ pathway, and LigD is its ligase.
- node: PANTHER:PTN000782634
  terms: [GO:0016843 amine-lyase activity]
  annotations: [UniProtKB:P37527 PdxS, B. subtilis]
  verdict: TRUE_BUT_GENERAL
  reason: Pyridoxal 5'-phosphate synthase is EC 4.3.3.6, an amine-lyase, so the term is true but
    general.
- node: PANTHER:PTN009136753
  terms: [GO:0030007 intracellular potassium ion homeostasis]
  annotations: [UniProtKB:P76007 NhaP2 E. coli, UniProtKB:P65524 S. typhimurium]
  verdict: FLAG_FAILS
  reason: Deleting the gene (ycgO, later cvrA) makes E. coli lose about half its intracellular K+ at
    low osmolarity, so the protein takes part in potassium homeostasis.
  supported_by:
  - reference_id: PMID:11700351
    supporting_text: Measurement of cytoplasmic alkali cations showed that an abrupt loss of about one-half
      of the intracellular K+ preceded the pause.
- node: PANTHER:PTN001092783
  terms: [GO:0000920 septum digestion after cytokinesis]
  annotations: [UniProtKB:P37690 EnvC E. coli, UniProtKB:Q8ZL55 S. typhimurium]
  verdict: FLAG_FAILS
  reason: EnvC activates the AmiA/AmiB amidases that cleave septal murein to separate daughter cells
    (UniProt FUNCTION); the term's definition covers that, and its constraints exclude only Metazoa
    and Amoebozoa.
- node: PANTHER:PTN002376512
  terms: [GO:0051787 misfolded protein binding]
  annotations: [UniProtKB:P47248 MG002 DnaJ-like, M. genitalium]
  verdict: UNCERTAIN
  reason: DnaJ-family co-chaperones bind unfolded and misfolded client proteins, so the term is
    plausible for MG002, but nothing shown here tests it.