Enrich causal graphs for the 6 single-edge DIET/syntrophy records

[realmarcin]

Completes the pending item from #253: all six 2-node records that carried one hand-authored donor→acceptor edge now have an Edison causal-graph pass and conservative curation.

Edges added

Record	Change
000070 Syntrophomonas_Methanospirillum	Reverse feedback edge (H2 scavenging → enables butyrate β-oxidation)
000069 Syntrophobacter_Methanospirillum	Reverse feedback edge (axenic-vs-syntrophic contrast, exact pair)
000068 Syntrophobacter_Methanobacterium	Reverse feedback edge + evidence correction (below)
000033 Geobacter_Methanosarcina	New T6SS-Associated Delay of DIET Establishment interaction + NEGATIVE edge

For 000070 the strongest evidence turned out to be in the already-cached abstract rather than the Edison report: PMID:16345745 gives both the dependency ("Growth and degradation of fatty acids occur only in syntrophic association with H(2)-using bacteria") and the perturbation ("The addition of H(2) … stopped growth and butyrate degradation"). No new reference needed.

The 000033 T6SS edge is curated as T6SS-associated, not T6SS-caused — the Hcp-deficient mutant is pleiotropic (it also reduces Fe(III) oxide faster than wild type, a phenotype not expected from losing the T6SS).

Evidence correction (000068)

Two evidence items quoted the Harmsen passage naming Methanospirillum hungateii but were explained as establishing Methanobacterium formicicum — this record's actual partner. Both are downgraded to PARTIAL with accurate explanations, exact-pair support is added from PMID:29611893, and the residual gap is filed as a discussion.

Where no edge was added

• 000031 — filed a CONTROVERSY discussion instead. ⚠️ This record's core framing is contested: the follow-up study (PMID:34939136) concludes the interaction is mediated (putative cobamide) and reports that 0.22-µm-filtered cell-free spent medium reproduces the metabolic shift, which would make pili contact unnecessary. Whether to re-scope the record away from community_category: DIET is left as a curator decision, not applied unilaterally.
• 000032 — an acetate cross-feeding node was proposed, but its quotations came from secondary reviews; the primary (doi:10.1039/C3EE42189A) is cached abstract-only and silent on acetate. Filed as a KNOWLEDGE_GAP per the 000176 precedent.

Verification

• just validate clean on all 6 records; network-integrity audit clean for all 6.
• Snippet audit: MATCH 4086 → 4101 (all 12 new snippets match), zero new mismatches.
• Caching PMID:29611893 OA full text additionally cleared the 3 pre-existing 000068 Methods-snippet mismatches (169 → 166 repo-wide).
• New reference caches: PMID:34939136, PMID:37650614.

Follow-ups surfaced

1. 000031 re-scoping decision — the highest-value open item.
2. Li et al. 2024 (doi:10.3390/w16243551) has exact-pair graded-formate perturbation data for 000068, but is not ingested: not in PubMed, no PMC record, so scripts/cache_fulltext.py can't verify its snippets. Would need a DOI-based full-text cache path.
3. just validate-references is a no-op — it reports Total checks: 0 even on untouched files. scripts/evidence_snippet_audit.py is what actually validates snippets. Worth fixing or documenting, since the recipe implies coverage it isn't providing.

🤖 Generated with Claude Code

[Copilot]

Pull request overview

This PR enriches the curated causal graphs for six previously “single-edge” DIET/syntrophy community records by adding conservative feedback edges and mechanistic/discussion annotations, alongside caching newly used references for reproducible evidence/snippet validation.

Changes:

• Added reciprocal/feedback edges for syntrophic partner scavenging (000068/000069/000070) and curated evidence to distinguish exact-pair vs imported support.
• Added a new negative, T6SS-associated “delay of DIET establishment” interaction for Geobacter–Methanosarcina (000033) and filed curated discussions instead of speculative edges for 000031/000032.
• Cached new references (including OA full text where available) to support snippet verification and reproducibility.

Reviewed changes

Copilot reviewed 10 out of 10 changed files in this pull request and generated 1 comment.

Show a summary per file

File	Description
references_cache/PMID_37650614.txt	Adds cached OA full text for the T6SS/DIET delay paper used in 000033 curation.
references_cache/PMID_34939136.txt	Adds cached abstract text for the follow-up mechanism paper informing the 000031 controversy discussion.
references_cache/PMID_29611893.md	Extends cached content to include OA full text supporting exact-pair evidence for 000068.
NEXT_TASKS.md	Marks the previously pending “single-edge enrichment” task as done and records follow-ups discovered.
kb/communities/Syntrophomonas_Methanospirillum_Syntrophy.yaml	Adds reciprocal (feedback) downstream edge and evidence supporting methanogen-side H₂ scavenging enabling butyrate oxidation.
kb/communities/Syntrophobacter_Methanospirillum_Syntrophy.yaml	Adds reciprocal (feedback) downstream edge and ties it to axenic-vs-syntrophic contrast evidence.
kb/communities/Syntrophobacter_Methanobacterium_Syntrophy.yaml	Corrects partner-attribution strength, adds exact-pair evidence, adds feedback edge, and adds a knowledge-gap discussion.
kb/communities/Geobacter_Methanosarcina_DIET.yaml	Adds T6SS-associated negative interaction impacting DIET establishment with supporting evidence.
kb/communities/Geobacter_Methanosaeta_DIET.yaml	Adds a knowledge-gap discussion documenting why an acetate cross-feeding edge was not curated.
kb/communities/Geobacter_Clostridium_DIET.yaml	Adds a controversy discussion capturing the contested contact-dependence vs mediated-mechanism interpretation.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.