# InterPro -> GO mapping review (SSSOM) -- proposed edits to interpro2go # # These rows review the InterPro2GO (https://www.ebi.ac.uk/GOA/InterPro2GO, GO_REF:0000002) mappings # of the InterPro entries deep-researched in projects/INTERPRO (see INTERPRO.md and the per-entry # interpro/interpro//-deep-research-falcon.md reports). Each subject InterPro entry carries # the listed GO term via interpro2go; this set records, per (entry, GO term), whether that mapping is # sound, over-broad, or should be removed. # # Three predicate classes encode the verdict: # * skos:exactMatch -- the GO term holds for essentially ALL proteins the entry # matches; the interpro2go mapping is sound (ACCEPT). # * skos:broadMatch -- the mapping is retained but the GO term is broader than / # over-annotates part of the entry's membership; recommend # demoting to a catalytically/functionally homogeneous child # entry or treating as non-core at the gene level # (MODIFY / KEEP_AS_NON_CORE / MARK_AS_OVER_ANNOTATED). # * skos:exactMatch + predicate_modifier: Not # -- the mapping is factually incorrect for the entry and is # proposed for REMOVAL from interpro2go (REMOVE). # # Provenance: every subject InterPro id, entry name, type, and GO-term set is from the cached # interpro/interpro//-metadata.yaml (InterPro API); every verdict + rationale is from the # corresponding falcon/Edison family deep-research report. GO ids/labels are validated non-obsolete # against the GO oak config. Validate with: just validate-interpro-mappings # (linkml-validate against the SSSOM schema 'mapping set' class, then GO-label term validation of the # generated interpro2go.terms.yaml). mapping_set_id: https://w3id.org/ai4curation/ai-gene-review/mappings/interpro2go-review mapping_set_title: InterPro2GO mapping review (deep-researched families) mapping_set_description: >- Curated review of the InterPro2GO mappings for the InterPro entries deep-researched in projects/INTERPRO. exactMatch rows endorse the interpro2go mapping; broadMatch rows retain it but flag it as over-broad for a structurally/functionally heterogeneous entry (demote to child entries or treat as non-core); exactMatch rows with predicate_modifier Not propose removing a factually incorrect mapping. Each verdict is backed by a family-level deep-research report (interpro/interpro//-deep-research-falcon.md). Seeds recommendations to InterPro2GO curators and quality-checks GO annotation of proteins matched by these signatures. license: https://creativecommons.org/licenses/by/4.0/ creator_label: - AI Gene Review project mapping_date: "2026-06-20" subject_source: InterPro object_source: GO curie_map: InterPro: https://www.ebi.ac.uk/interpro/entry/InterPro/ GO: http://purl.obolibrary.org/obo/GO_ skos: http://www.w3.org/2004/02/skos/core# semapv: https://w3id.org/semapv/vocab/ sssom: https://w3id.org/sssom/ mappings: # ===== IPR000719 Protein kinase domain (type: domain) -- REMOVE both (pseudokinases) ===== - subject_id: InterPro:IPR000719 subject_label: Protein kinase domain predicate_id: skos:exactMatch predicate_label: exact match predicate_modifier: Not object_id: GO:0005524 object_label: ATP binding mapping_justification: semapv:ManualMappingCuration comment: >- REMOVE. IPR000719 is a domain signature that also matches pseudokinases; Class I pseudokinases bind neither ATP nor metal, so ATP binding is not universal across the entry. Restrict to catalytic child entries with intact ATP-binding motifs. (IPR000719 deep research.) - subject_id: InterPro:IPR000719 subject_label: Protein kinase domain predicate_id: skos:exactMatch predicate_label: exact match predicate_modifier: Not object_id: GO:0006468 object_label: protein phosphorylation mapping_justification: semapv:ManualMappingCuration comment: >- REMOVE at the domain level. Pseudokinases (~10% of the human kinome) lack catalytic residues and do not phosphorylate; some are repurposed (AMPylation, glutamylation). Restrict to catalytic child families and use specific MF terms (GO:0004674 / GO:0004713) rather than this BP term on a domain. # ===== IPR001128 Cytochrome P450 (type: family) -- cofactor binding sound; activity over-broad ===== - subject_id: InterPro:IPR001128 subject_label: Cytochrome P450 predicate_id: skos:exactMatch predicate_label: exact match object_id: GO:0020037 object_label: heme binding mapping_justification: semapv:ManualMappingCuration comment: >- ACCEPT. Heme (heme-thiolate) coordination is the defining, universally conserved feature of the P450 fold. (IPR001128 deep research.) - subject_id: InterPro:IPR001128 subject_label: Cytochrome P450 predicate_id: skos:exactMatch predicate_label: exact match object_id: GO:0005506 object_label: iron ion binding mapping_justification: semapv:ManualMappingCuration comment: ACCEPT. The central heme iron is conserved across all P450s. - subject_id: InterPro:IPR001128 subject_label: Cytochrome P450 predicate_id: skos:broadMatch predicate_label: broad match object_id: GO:0004497 object_label: monooxygenase activity mapping_justification: semapv:ManualMappingCuration comment: >- Over-broad. The family spans 819+ functionally diverse subfamilies (incl. non-monooxygenase activities and catalytically inert members); monooxygenase activity over-annotates at the family level. Demote to functionally homogeneous child entries. - subject_id: InterPro:IPR001128 subject_label: Cytochrome P450 predicate_id: skos:broadMatch predicate_label: broad match object_id: GO:0016705 object_label: oxidoreductase activity, acting on paired donors, with incorporation or reduction of molecular oxygen mapping_justification: semapv:ManualMappingCuration comment: Over-broad for the same reason as monooxygenase activity; demote to child entries. # ===== IPR001424 Cu/Zn SOD domain (type: domain) ===== - subject_id: InterPro:IPR001424 subject_label: Superoxide dismutase, copper/zinc binding domain predicate_id: skos:broadMatch predicate_label: broad match object_id: GO:0046872 object_label: metal ion binding mapping_justification: semapv:ManualMappingCuration comment: >- KEEP_AS_NON_CORE. Generic and weakly informative; domain presence does not guarantee canonical Cu/Zn binding. (IPR001424 deep research.) - subject_id: InterPro:IPR001424 subject_label: Superoxide dismutase, copper/zinc binding domain predicate_id: skos:exactMatch predicate_label: exact match predicate_modifier: Not object_id: GO:0006801 object_label: superoxide metabolic process mapping_justification: semapv:ManualMappingCuration comment: >- REMOVE from this domain-level entry. A BP term on a structural module; copper-only SODs and copper-chaperone members (e.g. CCS) carry the domain but do not dismutate superoxide, so the process term mis-annotates divergent members. Map process terms only onto bona fide SOD child entries. # ===== IPR000276 GPCR, rhodopsin-like (type: family) ===== - subject_id: InterPro:IPR000276 subject_label: G protein-coupled receptor, rhodopsin-like predicate_id: skos:broadMatch predicate_label: broad match object_id: GO:0004930 object_label: G protein-coupled receptor activity mapping_justification: semapv:ManualMappingCuration comment: >- Over-broad. Atypical chemokine receptors (ACKRs) and some orphan receptors carry the Class A signature but lack canonical G-protein coupling (signal via arrestin), so GPCR activity over-annotates these exceptions. (IPR000276 deep research.) - subject_id: InterPro:IPR000276 subject_label: G protein-coupled receptor, rhodopsin-like predicate_id: skos:broadMatch predicate_label: broad match object_id: GO:0007186 object_label: G protein-coupled receptor signaling pathway mapping_justification: semapv:ManualMappingCuration comment: Over-broad for the same ACKR/orphan exceptions as GPCR activity. - subject_id: InterPro:IPR000276 subject_label: G protein-coupled receptor, rhodopsin-like predicate_id: skos:broadMatch predicate_label: broad match object_id: GO:0016020 object_label: membrane mapping_justification: semapv:ManualMappingCuration comment: KEEP_AS_NON_CORE. Universally true but generic; uninformative as a core annotation. # ===== IPR001046 NRAMP / SLC11 metal transporter (type: family) -- broad terms sound ===== - subject_id: InterPro:IPR001046 subject_label: NRAMP family predicate_id: skos:exactMatch predicate_label: exact match object_id: GO:0046873 object_label: metal ion transmembrane transporter activity mapping_justification: semapv:ManualMappingCuration comment: >- ACCEPT as a broad family-level term. All members are integral-membrane transition-metal transporters; do not add more specific terms (e.g. iron transport) at this level, as substrate specificity diverges by subfamily. (IPR001046 deep research.) - subject_id: InterPro:IPR001046 subject_label: NRAMP family predicate_id: skos:exactMatch predicate_label: exact match object_id: GO:0030001 object_label: metal ion transport mapping_justification: semapv:ManualMappingCuration comment: ACCEPT as a broad family-level process term. - subject_id: InterPro:IPR001046 subject_label: NRAMP family predicate_id: skos:broadMatch predicate_label: broad match object_id: GO:0016020 object_label: membrane mapping_justification: semapv:ManualMappingCuration comment: KEEP_AS_NON_CORE. Trivially true but informationally useless as a core annotation. # ===== IPR012724 Chaperone DnaJ (type: family) ===== - subject_id: InterPro:IPR012724 subject_label: Chaperone DnaJ predicate_id: skos:exactMatch predicate_label: exact match predicate_modifier: Not object_id: GO:0005524 object_label: ATP binding mapping_justification: semapv:ManualMappingCuration comment: >- REMOVE. Factually incorrect: ATP binding/hydrolysis is a property of the Hsp70 partner; J-domain proteins use the HPD motif to stimulate Hsp70 ATPase and lack an ATP-binding pocket. No evidence supports universal ATP binding by DnaJ family members. (IPR012724 deep research.) - subject_id: InterPro:IPR012724 subject_label: Chaperone DnaJ predicate_id: skos:exactMatch predicate_label: exact match object_id: GO:0006457 object_label: protein folding mapping_justification: semapv:ManualMappingCuration comment: ACCEPT. J-domain proteins are co-chaperones that promote client folding; broadly appropriate. - subject_id: InterPro:IPR012724 subject_label: Chaperone DnaJ predicate_id: skos:broadMatch predicate_label: broad match object_id: GO:0009408 object_label: response to heat mapping_justification: semapv:ManualMappingCuration comment: >- KEEP_AS_NON_CORE / demote. Overgeneralizes from heat-inducible canonical members to a functionally heterogeneous family (many JDPs are constitutive or organelle-specific). Restrict to stress- responsive subfamilies. # ===== IPR007197 Radical SAM (type: domain) -- both sound despite genericity ===== - subject_id: InterPro:IPR007197 subject_label: Radical SAM predicate_id: skos:exactMatch predicate_label: exact match object_id: GO:0003824 object_label: catalytic activity mapping_justification: semapv:ManualMappingCuration comment: >- ACCEPT. Notable case: this is the root MF term and maximally generic, yet it is the CORRECT family-level annotation. Radical SAM members catalyze mechanistically diverse reactions (>100 families), so no specific catalytic term is universal; the conserved property really is just "is a catalytic enzyme". Replacing with e.g. oxidoreductase activity would over-annotate. Optionally KEEP_AS_NON_CORE. (IPR007197 deep research.) - subject_id: InterPro:IPR007197 subject_label: Radical SAM predicate_id: skos:exactMatch predicate_label: exact match object_id: GO:0051536 object_label: iron-sulfur cluster binding mapping_justification: semapv:ManualMappingCuration comment: >- ACCEPT. The [4Fe-4S] cluster that reductively cleaves SAM is the defining, universally conserved cofactor of the domain; substantially more informative than generic metal ion binding. # ===== IPR020849 Small GTPase, Ras-type (type: family) ===== - subject_id: InterPro:IPR020849 subject_label: Small GTPase, Ras-type predicate_id: skos:exactMatch predicate_label: exact match object_id: GO:0005525 object_label: GTP binding mapping_justification: semapv:ManualMappingCuration comment: ACCEPT. The universal, defining molecular function of the family. (IPR020849 deep research.) - subject_id: InterPro:IPR020849 subject_label: Small GTPase, Ras-type predicate_id: skos:exactMatch predicate_label: exact match object_id: GO:0003924 object_label: GTPase activity mapping_justification: semapv:ManualMappingCuration comment: >- PROPOSED ADD -- absent from interpro2go (annotation gain). The GTP-hydrolysis machinery (P-loop, Switch II catalytic Gln61, Mg2+ coordination) is described as universal and essential across Ras-type small GTPases; the entry maps GTP binding but not the catalytic hydrolysis it defines. Intrinsic hydrolysis is slow and GAP-accelerated, so flag for curator confirmation before adding. - subject_id: InterPro:IPR020849 subject_label: Small GTPase, Ras-type predicate_id: skos:broadMatch predicate_label: broad match object_id: GO:0007165 object_label: signal transduction mapping_justification: semapv:ManualMappingCuration comment: >- Over-broad. True but too generic; assign specific child terms (e.g. GO:0007265 Ras protein signal transduction) at subfamily entries with pathway evidence rather than this broad BP term family-wide. - subject_id: InterPro:IPR020849 subject_label: Small GTPase, Ras-type predicate_id: skos:broadMatch predicate_label: broad match object_id: GO:0016020 object_label: membrane mapping_justification: semapv:ManualMappingCuration comment: >- MARK_AS_OVER_ANNOTATED. Generic; specific membrane-compartment terms need subfamily evidence, and not all Ras-type members are membrane-associated. Assign compartment terms to child entries instead. # ===== IPR002100 Transcription factor, MADS-box (type: domain) -- both sound ===== - subject_id: InterPro:IPR002100 subject_label: Transcription factor, MADS-box predicate_id: skos:exactMatch predicate_label: exact match object_id: GO:0003677 object_label: DNA binding mapping_justification: semapv:ManualMappingCuration comment: >- ACCEPT. Sequence-specific DNA binding is intrinsic to the MADS-box domain. Note: do NOT add DNA-binding transcription factor activity (GO:0003700) here -- the report finds TF/regulatory function is a WHOLE-PROTEIN property arising from the K/C domains and complex context, not the MADS domain itself, so it would over-annotate the domain entry. (IPR002100 deep research.) - subject_id: InterPro:IPR002100 subject_label: Transcription factor, MADS-box predicate_id: skos:exactMatch predicate_label: exact match object_id: GO:0046983 object_label: protein dimerization activity mapping_justification: semapv:ManualMappingCuration comment: ACCEPT. MADS-box proteins bind DNA as dimers; dimerization is domain-intrinsic.