InterPro mapping review — historical notes

Historical session notes for the InterPro mapping review. These preserve the original chronology and provisional interpretations; consult the project page and linked mapping set for the current summary.

2026-09-27 — rotary-ATPase mappings from the FliI/SctN audit

The TreeGrafter rotary-ATPase leak audit reviewed
nine flagellar (FliI) and injectisome (SctN) export ATPases, and three InterPro2GO sources
turned up alongside the PAINT/TreeGrafter error. Seven rows were added to the
mapping set (29 -> 36 mappings, twelve -> fifteen entries). Match
counts are from QuickGO withFrom queries on 2026-09-27.

• IPR013380 -> GO:0046961, GO:0006754 — REMOVE. The entry is specific to SctN,
  so a wrong term is wrong for all 915 annotated members rather than for an exception
  subset. Both terms read SctN as an F1 synthase. The process and complex mappings are
  endorsed.
• IPR004100 -> GO:1902600, GO:0046034 — NARROW. A domain shared by genuine
  rotary ATPase subunits and the FliI/SctN paralogs (71,788 annotations in all). The fix
  is conditional suppression when IPR005714 also matches, not removal.
• IPR005714 -> GO:0009058 — REMOVE. Present in the 2025-09 mapping file but with
  no annotations in QuickGO, so it may already be filtered downstream.

2026-09-17 — three mappings found gene-first rather than family-first

The contested-function review (Contested gene functions, 2025-2026)
read ~60 human genes one at a time and, as a side effect, caught three InterPro2GO
mappings the family worklist had not reached. They are in the
mapping set (25 -> 29 mappings, nine -> twelve entries) and
summarized on the project page.
Entry names, types and match counts were verified against the live InterPro API and the
GO term definitions against QuickGO.

• IPR045122 -> GO:0005227 — REMOVE. A name collision, distinct from the
  fold-is-not-function pattern that drives the rest of this worklist: the entry is
  "Calcium permeable stress-gated cation channel 1-like" and the GO term means a channel
  that opens when calcium binds it. The two readings of "calcium ... cation channel"
  are not distinguishable without reading the term definition. Lands on TMEM63A/B/C.
• IPR042371 -> GO:0003726 — REMOVE. A catalytic activity mapped onto a binding
  domain: the deaminase is in ADAR1's separate domain, so ZBP1 — a Zalpha domain with no
  deaminase domain — inherits an activity it cannot perform. The entry's GO:0003723
  RNA binding is sound and is recorded as an ACCEPT row.
• IPR006935 -> GO:0003677 — REMOVE. The "Helicase/UvrB, N-terminal" entry is
  dominated by DNA-acting enzymes, but the fold recurs in the RNA-sensing RIG-I-like
  receptors, so IFIH1 (MDA5), a cytosolic dsRNA sensor, inherits DNA binding. The
  nucleic-acid substrate is a property of the enzyme, not of the shared fold;
  GO:0005524 and GO:0016787 on the same entry are unaffected.

The methodological point is about intake, not about these three entries. The family
worklist is ranked by how many reviewers already flagged an entry, so it is
structurally blind to mappings that are wrong but rarely reviewed — IPR045122 has
three human members and would never rise up that ranking. Harvesting InterPro defects
out of deep gene reviews is a complementary intake path, and close to free when the
gene review is happening anyway.

Open follow-up

These unfinished tasks are carried forward from the recorded workstreams and research
notes. Their completion has not been reassessed as part of the page reorganization.

• [ ] Re-run the deferred sigma-54 family research and verify that the report contains
  retrieved evidence and real citations before using it for a mapping proposal.
• [ ] Re-run the deferred pseudouridine synthase family research with the same evidence check.
• [ ] Re-run the deferred GAPDH family research with the same evidence check.
• [ ] Continue down the entry worklist: confirm flagged
  gene-level verdicts, assess canonical replacement terms, and update gene reviews.
• [ ] Check accepted annotations on exception members identified by family research.
• [ ] Summarize per-entry recommendations for InterPro2GO curators.

Archived status — 2026-06-20

The original checklist and verdict table are preserved below as a historical snapshot.
Counts and provisional interpretations reflect that session; the
project findings and
mapping set provide the current summary.

Workstream 1 — review flagged mappings

• [x] Extractor that aggregates InterPro2GO verdicts across all reviews
• [x] Per-entry priority worklist (1826 entries; 265 with ≥3 suspect mappings)
• [ ] Work down the top entries: confirm the suspect verdicts, propose canonical
  replacement terms, and feed corrections back to the gene reviews
• [ ] Summarize per-entry recommendations for InterPro2GO curators

Workstream 2 — family deep research

• [x] Confirmed the metadata fetcher supports interpro (IPR) entries
• [x] Generated deep-research process: templates/interpro_family_research.md,
  scripts/deep_research_interpro_family.py, and the
  just deep-research-interpro-family recipe
• [x] Seed example cached (interpro/interpro/IPR000719/)
• [x] First family deep research generated with falcon/Edison
  (IPR000719-deep-research-falcon.md): verdict that both InterPro2GO terms
  (ATP binding, protein phosphorylation) over-annotate the domain because it
  also matches pseudokinases — REMOVE at the domain level, restrict to catalytic
  children (GO:0004674 / GO:0004713)
• [x] Batch 1 of 5 more top entries researched with falcon/Edison (P450, Cu/Zn SOD,
  GPCR, NRAMP/SLC11, DnaJ) — see the table below
• [x] Began feeding verdicts back into gene reviews (DnaJ family): the reviews are
  strongly concordant with the family research — all 7 DnaJ genes with the
  InterPro2GO ATP binding annotation already flagged it (5 REMOVE, 1 MODIFY-to
  ATPase-activator, 1 over-annotated). Hardened the one soft outlier (yeast/YDJ1,
  MARK_AS_OVER_ANNOTATED → REMOVE) and attached the IPR012724 family report as
  corroborating evidence.
• [ ] Per-family, hunt for the higher-value case: a gene that currently ACCEPTs the
  flagged term but is actually one of the verdict's exception members (pseudokinase,
  copper chaperone, atypical chemokine/orphan receptor, non-catalytic P450) — a genuine
  missed over-annotation rather than a soft-vs-hard mismatch
• [x] Captured the family verdicts as proposed interpro2go edits in SSSOM YAML
  (INTERPRO/interpro2go.sssom.yaml, 17 mappings over the 6 entries) — the
  consortium-facing deliverable, validated via just validate-interpro-mappings
• [ ] Continue down the worklist (interpro_family_priorities.tsv)

Family deep-research verdicts (falcon/Edison)

InterPro	Family	Entry type	InterPro2GO verdict
IPR000719	Protein kinase domain	domain	ATP binding + protein phosphorylation → REMOVE at domain level (captures pseudokinases); restrict to catalytic children (GO:0004674 / GO:0004713)
IPR001128	Cytochrome P450	family	heme binding + iron ion binding universal → keep; monooxygenase activity + oxidoreductase activity over-annotate (819+ functionally diverse families)
IPR001424	Cu/Zn superoxide dismutase domain	domain	superoxide metabolic process → REMOVE (BP term on a structural module; copper-chaperone members don't dismutate); metal ion binding → KEEP_AS_NON_CORE
IPR000276	GPCR, rhodopsin-like (Class A)	family	GPCR activity + GPCR signaling pathway → MARK_AS_OVER_ANNOTATED / MODIFY (atypical chemokine + orphan receptors lack canonical G-protein coupling); membrane → KEEP_AS_NON_CORE
IPR001046	NRAMP / SLC11 metal transporter	family	metal ion transmembrane transporter activity + metal ion transport → ACCEPT as broad family terms; membrane → KEEP_AS_NON_CORE; do not add more specific terms at family level
IPR012724	Chaperone DnaJ (J-domain)	family	ATP binding → REMOVE (factually wrong — the Hsp70 partner binds ATP, not DnaJ); protein folding → ACCEPT; response to heat → KEEP_AS_NON_CORE (only heat-inducible subfamilies)
IPR007197	Radical SAM	domain	catalytic activity → ACCEPT despite being the MF root term — see note below; iron-sulfur cluster binding → ACCEPT (defining [4Fe-4S] cofactor)
IPR020849	Small GTPase, Ras-type	family	GTP binding → ACCEPT; ADD GTPase activity (GO:0003924) — proposed new mapping (annotation gain); signal transduction → demote to subfamily (GO:0007265); membrane → MARK_AS_OVER_ANNOTATED
IPR002100	Transcription factor, MADS-box	domain	DNA binding + protein dimerization activity → ACCEPT (both domain-intrinsic). Notably do NOT add DNA-binding TF activity — TF function is a whole-protein property (K/C domains + complex), not the MADS domain

• [ ] Run just deep-research-interpro-family <IPR> (falcon/Edison default) for the next entries

Last updated: 2026-06-20

Session notes — 2026-06-20

Project creation. Scoped the InterPro2GO (GO_REF:0000002) review. Built the
extractor and the per-entry priority worklist from all 2732 reviewed genes: 3652
InterPro2GO annotations, 47% flagged suspect across 1826 InterPro entries. Broad
domain/superfamily signatures dominate the suspect list (protein kinase domain, P450,
Cu/Zn SOD, GPCR), confirming the "fold ≠ function" failure mode as the main driver.

Closed the PANTHER-vs-InterPro deep-research gap. Gene deep research is a generated
process (just deep-research-<provider>); there was no equivalent for the InterPro
entries behind InterPro2GO annotations. Added the InterPro-family analogue —
templates/interpro_family_research.md, scripts/deep_research_interpro_family.py, and
the just deep-research-interpro-family <IPR> [provider] recipe (provider defaults to
falcon/Edison) — so families are researched by the same generated pipeline (output:
interpro/<db>/<ID>/<ID>-deep-research-<provider>.md), with IPR000719 cached as a
seed.

Batch 2 + first proposed new mapping. Researched 6 more families; 3 grounded cleanly
(Radical SAM, Ras-type small GTPase, MADS-box) and are in the SSSOM (now 25 mappings).
Notable findings:

• Genericity ≠ wrongness (Radical SAM, IPR007197). I predicted the MF root term
  catalytic activity would be a REMOVE. The research says ACCEPT: because the
  superfamily catalyzes >100 mechanistically different reactions, the only universally
  true MF really is "is an enzyme", so the maximally generic term is the correct
  family-level annotation — replacing it with anything more specific would over-annotate.
• First annotation-gain proposal (Ras-type, IPR020849). The entry maps GTP binding
  but not GTPase activity (GO:0003924), even though the GTP-hydrolysis machinery
  (P-loop, Switch II/Gln61, Mg²⁺) is universal — so we propose ADDING it (an
  exactMatch row, flagged for curator confirmation since intrinsic hydrolysis is
  GAP-accelerated).
• Domain-intrinsic vs whole-protein (MADS-box, IPR002100). I expected to add
  DNA-binding TF activity; the research argues against it — the MADS domain provides
  DNA binding + dimerization, but being a transcription factor is a whole-protein
  property (K/C domains, complex context), so adding it would over-annotate the domain.
• QC catch. 3 of the 6 runs (sigma-54, pseudouridine synthase, GAPDH) silently
  returned ungrounded reports (exit 0, file written, but "no contexts were retrieved
  … not grounded in evidence", zero real citations) — likely Edison retrieval throttling
  under 6-way parallel load. A sequential re-run also failed fast (~3 s each, no
  retrieval), so this is a transient Edison retrieval-backend outage, not a load issue —
  these 3 entries are deferred (metadata cached) for re-running when the backend
  recovers. Excluded from the SSSOM. (Detect with: grep for "no contexts were retrieved"
  or citation_count/zero real refs — a groundedness guard worth adding to the wrapper.)

Batch 1 of family deep research (falcon/Edison). Ran five more top entries: P450
(IPR001128), Cu/Zn SOD (IPR001424), GPCR Class A (IPR000276), NRAMP/SLC11
(IPR001046), and DnaJ (IPR012724). See the archived family verdict table. A
recurring, independently-reached pattern: cofactor/binding terms (heme binding, metal ion binding) and broad transport terms hold family-wide, but whole-protein activity
and process terms attached to a structural module over-annotate — most sharply for
IPR012724, where Edison flags ATP binding as factually wrong on DnaJ (the Hsp70 partner
binds ATP), and for IPR001424, where superoxide metabolic process mis-annotates
copper-chaperone members that do not dismutate.