Add Gao-Her-DUF system

[realmarcin]

Summary

Adds traitmech:000361 for the Gao-Her-DUF phage-defense system as a GENOMICS / CLASS / PROPOSED TraitRecord.

The record interprets Gao_Her_DUF as a genome-level DefenseFinder Gao-family system possession trait and keeps the Gao_Her_DUF__DUF4297 and Gao_Her_DUF__HerA_DUF HMM profile rows as RELATED_SYNONYM provenance/evidence, not individual protein traits.

Identity and Evidence

• label: Gao-Her-DUF system
• identifier: traitmech:000361
• parent: traitmech:000209 phage defense system
• upstream placeholder: METPO:1031500 in proposals/metpo_traitmech_v238
• primary DOI: DOI:10.1126/science.aba0372
• PMID: PMID:32855333
• pinned stable registry sources:
  • List_system_article.md at afb0e5a8b466be53586b13266f5d38d98c3ac268
  • DefenseFinder_rules.tsv at afb0e5a8b466be53586b13266f5d38d98c3ac268
  • Liste_hmm_system.md at afb0e5a8b466be53586b13266f5d38d98c3ac268

Curation

• adds data/traits/genomics/gao_her_duf_system.yaml
• adds history/records/gao_her_duf_system/2026-09-22T093443Z-codex-da0d31.yaml
• adds scripts/add_gao_her_duf_system_trait.py
• reserves METPO:1031500 in proposals/metpo_traitmech_v238
• updates POST_444_CANONICAL_EXAMPLE_QUEUE
• regenerates pages, the priority dashboard, discussion browser data, and graph/proposal audit reports

Validation

• .venv/bin/linkml-validate -s src/traitmech/schema/traitmech.yaml --target-class TraitRecord data/traits/genomics/gao_her_duf_system.yaml
• .venv/bin/python scripts/validate_strict.py data/traits/genomics/gao_her_duf_system.yaml
• .venv/bin/python scripts/verify_metpo_proposal.py proposals/metpo_traitmech_v238
• .venv/bin/python scripts/validate_history_links.py history/records/gao_her_duf_system
• .venv/bin/python scripts/backfill_canonical_examples_444.py
• .venv/bin/python scripts/render_trait_pages.py
• .venv/bin/python scripts/trait_priority.py --dashboard --top 80
• PYTHONPATH=/Users/marcin/Documents/VIMSS/ontology/KG-Hub/KG-Microbe/culturebotai-claw/src .venv/bin/python -m kg_microbe_discussions --config conf/discussions_config.yaml --output app/discussions
• .venv/bin/python scripts/audit_graph_protein_taxa.py --fail-on gaps
• .venv/bin/python scripts/ground_causal_nodes.py
• .venv/bin/python scripts/audit_proposals.py
• .venv/bin/python scripts/audit_evidence_snippets.py
• .venv/bin/python scripts/audit_schema.py
• .venv/bin/python scripts/audit_writers.py
• .venv/bin/python scripts/verify_metpo_proposal.py --coverage
• .venv/bin/python scripts/audit_causal_graphs.py
• .venv/bin/python scripts/ground_causal_predicates.py
• .venv/bin/python scripts/audit_biolink_curies.py
• .venv/bin/python scripts/audit_predicate_domains.py --fail-on new
• .venv/bin/python scripts/audit_discussions_data.py
• .venv/bin/python scripts/audit_canonical_examples.py --no-resolve
• .venv/bin/python scripts/check_biolink_coverage.py
• .venv/bin/python scripts/audit_exact_synonyms.py --collisions-only
• .venv/bin/python scripts/audit_unapplied_groundings.py
• .venv/bin/python scripts/audit_discussion_anchors.py
• .venv/bin/python scripts/pr_sanity.py
• .venv/bin/python scripts/audit_justfile_paths.py
• .venv/bin/python scripts/audit_qc_paths_coverage.py
• .venv/bin/python scripts/run_trait_graph_audit.py --verify
• .venv/bin/python scripts/check_sources.py
• .venv/bin/python scripts/verify_snippets.py --delay 2 --record data/traits/genomics/gao_her_duf_system.yaml
• .venv/bin/ruff check src scripts tests
• .venv/bin/python -m pytest tests/test_readme_artifacts.py tests/test_trait_priority.py -v --tb=short
• .venv/bin/python -m pytest -q
• .venv/bin/python scripts/audit_history_records.py --base origin/main
• git diff --check
• git diff --cached --check

[culturebot-reviewer]

Approving. This is a faithful instance of the merged Gao-family pattern and I found no defects.

Checks that passed:

• Identity. traitmech:000361 is unique across data/traits/ and continues the sequence after traitmech:000360 (Gao-TerY, #1241). METPO:1031500 and metpo_traitmech_v238 are unused elsewhere. Parent traitmech:000209 resolves to data/traits/genomics/phage_defense_system.yaml, label "phage defense system" — matches the node grounding.
• Predicates. RO:0002326 = "contributes to" and rdfs:subClassOf = "is a" match mappings/predicate_grounding.tsv (both */* domain/range). METPO:2007700 "confers" is used with a BIOLOGICAL_PROCESS subject and a TRAIT object, both inside that row's gated type lists.
• Graph structure. All three edges reference declared nodes, and all four nodes sit on the connected path gao_her_duf_locus → gao_2020_antiviral_cassette_defense → gao_her_duf_system_trait → phage_defense_system. Nothing orphaned or unreachable from the trait node.
• Enums. RELATED_SYNONYM, NONMECHANISTIC, TRAIT, GENETIC_ELEMENT, BIOLOGICAL_PROCESS are all in traitmech.yaml; KNOWLEDGE_GAP is in mech_shared.yaml; CurationEvent.action is an unconstrained string.
• Derived artifacts. README's 756 total and 153 PROPOSED genomics both match live file counts. pages/traits/genomics/gao_her_duf_system.html, priority.json, discussions/data.js, and the three reports/*.tsv deltas are all present and consistent; the history record under history/records/gao_her_duf_system/ is there.

One correct detail worth naming, since it is where this record diverges from its siblings: the Gao_Her_DUF synonym is sourced to DefenseFinder_rules.tsv rather than List_system_article.md, and the article-registry evidence note says "the broader Gao_Her family key." That is the right call — Gao_Her_DUF is a subsystem key that does not appear in the article registry — and the record does not overclaim by listing Gao_Her itself as a synonym.

Scope note on my verification: this review environment has no network access, so I could not re-fetch the three pinned defense-finder-models files at afb0e5a8b4 or the PMID:32855333 abstract to confirm the snippets are verbatim. What I could check: the PMID sentence is byte-identical to the one carried by ten merged Gao-family records; the List_system_article.md row (including the escaped 10\\.1126/science\\.aba0372) matches those records exactly; and the two Liste_hmm_system.md rows reproduce the upstream fixed-width column padding correctly for the new, longer profile names (49/23/24/7 characters, same as the differently-sized TerY rows). That is strong but indirect evidence. The direct evidence remains your scripts/verify_snippets.py --record ... run.