Fusion coculture graph conflates wild-type and engineered assays and overstates causal mechanisms #1316

Finding

Source-based self-adversarial review of kb/communities/Clostridium_Acetobutylicum_Ljungdahlii_Fusion_Coculture.yaml (CommunityMech:000336) finds exact-system and causal-certainty errors in its nine-node/three-arrow graph.

• The record explicitly describes a wild-type pairing, but its DNA-integration node and fusion-to-DNA arrow use PMID:38214507, whose donor is plasmid-bearing C. ljungdahlii-ptaHalo and whose recipients underwent selection. DNA integration is supported in that engineered experiment, not demonstrated in this wild-type pair. Fusion is the paper's most-likely route, not isolated by a fusion-specific intervention.
• PMID:32873766 supports wild-type structural fusion and chemical protein-label exchange. Its RNA experiment uses a C. acetobutylicum-Halo reporter. The graph currently conflates these assays and wild-type DNA exchange.
• That source explicitly says C. ljungdahlii can in principle produce 2,3-butanediol alone, but did not in its controls. The current universal inability claim and definitive fusion-to-product mechanism are too strong.
• PMID:40298437 measures amino-acid profiles, supplementation responses and early glucose-consumption rates. These must be retained, but direct donor-to-recipient amino-acid flux, hydrogenase mediation, reciprocal growth from transcript profiles, and competition/product-inhibition causality remain proposed. Monocultures used different substrate/gas conditions, so their existence does not establish facultative cooperation under identical conditions.

Acceptance

Remove the engineered DNA node/arrow from the wild-type graph with a preserved ledger snapshot. Bound remaining claims to the actual experiments, qualify uncertain arrows and add focused knowledge gaps. Preserve the supported gas link, structural fusion and condition-specific products. Do not add nodes/arrows or change schema enums as part of this repair. Regenerate pages and pass focused schema, strict, term, reference, snippet, network, history and regression gates.

Primary main bodies, Methods and captions read for 2025 transcriptome, 2020 structure, and 2024 DNA. Founding PMID:30391511 is abstract-bounded. Images/supplements not visually inspected.

Deduplication: 659 all-state issue titles/bodies and ignored-inclusive local searches. Existing #749 and #752 track broader enum questions, not this exact-system repair. Non-graph metadata gets a separate followup.