id: CR-0007
title: OmcS-like extracellular multiheme cytochromes of metal-reducing bacteria bind heme
  and carry electrons
status: PROPOSED
description: >-
  Assigns heme binding and electron transfer activity to proteins matching the CDD model
  cd21555, "C-type cytochrome OmcS and similar proteins". The model is a hand-curated,
  tight signature for the OmcS branch of extracellular multiheme cytochromes, and its 56
  UniProtKB members are confined to metal-reducing lineages (Geobacteraceae, Anaeromyxobacter,
  Candidatus Sulfobium). It is not integrated into any InterPro entry and carries no GO of
  its own, so it reaches GOA through no route at all: 52 of its 56 members have no heme
  binding annotation despite every one of them being a c-type cytochrome. This is new
  coverage rather than a replacement; nothing currently fires on this model.
condition_sets:
- number: 1
  conditions:
  - condition_type: CDD
    value: cd21555
    label: OmcS-like
  notes: >-
    The CDD model alone. No taxon condition is added because it would not separate right
    from wrong members: every member already falls in a metal-reducing bacterial lineage,
    so a Bacteria constraint would restrict nothing. The model is specific enough on its
    own, matching 56 proteins of 341-477 residues, all annotated as c-type cytochromes or
    as doubled-CXXCH-domain proteins.
go_annotations:
- go_id: GO:0020037
  go_label: heme binding
  aspect: molecular_function
- go_id: GO:0009055
  go_label: electron transfer activity
  aspect: molecular_function
gene_reviews:
- genes/GEOSL/omcS/omcS-ai-review.yaml
validation_runs:
- run_date: '2026-11-10'
  uniprot_query: xref:cdd-cd21555
  matched_count: 56
  notes: >-
    Closure-aware gain measured with projects/NCBIFam/cdd_go_gap_probe.py on 2026-11-10,
    counting entries that lack the term and every descendant. GO:0020037 heme binding:
    56 carry the model, 4 already have the term, gain 52 (reviewed: 1 carries, 0 have,
    gain 1). GO:0009055 electron transfer activity: 53 of 56 already have it, gain 3
    (reviewed gain 0); it is included because it is the correct molecular function and
    completes the pair, not for the gain. Members span Geobacter, Geotalea, Geomonas,
    Citrifermentans, Geoalkalibacter, Geomobilimonas, Trichloromonas, Malonomonas,
    Anaeromyxobacter and Candidatus Sulfobium, and include G. sulfurreducens OmcT
    (UniProtKB:Q74A87) alongside OmcS.
supported_by:
- reference_id: PMID:21236241
  supporting_text: Purified OmcS had a molecular mass of 47015 Da, and six low-spin bis-histidinyl
    hexacoordinated heme groups.
- reference_id: PMID:30951668
  supporting_text: The inter-subunit interfaces show unique structural elements such as inter-subunit
    parallel-stacked hemes and axial coordination of heme by histidines from neighboring subunits.
notes: >-
  Three terms with a larger measured gain were considered and deliberately left out,
  because gain is not the test; safety across all members is.

  GO:0019645 anaerobic electron transport chain (gain 54) is a claim about how the
  organism respires, not about the protein. Every current member happens to sit in a
  metal-reducing lineage, so it is probably true of all of them, but "probably" is the
  standard this rule exists to avoid, and a rule fires on future members too.

  GO:0009986 cell surface (gain 55) is established for OmcS itself and is likely for the
  rest, but no member other than OmcS has been localized. None of these proteins carries a
  transmembrane segment or lipid anchor, so the attachment mode is not predictable from
  sequence either.

  GO:0051258 protein polymerization (gain 56) is the trap. OmcS polymerizes into a
  filament and so, separately, does OmcE, but that is demonstrated per protein by cryo-EM,
  not implied by membership of this model. Propagating it family-wide from one structure
  is exactly the over-propagation this rules directory exists to correct.

  Context: this model is one of 14,427 CDD models (72.5% of the 19,902 InterPro knows)
  that InterPro has not integrated, and CDD-proper carries no GO of its own, so none of
  them reaches GOA. See projects/NCBIFam.md. The partner finding is that cd21555 is a good
  case and an unrepresentative one: across a 40-model sample of unintegrated models, most
  matched proteins already carry a molecular function from another route.