Consistency:CONTRADICTION flagged. PN row 3 projects GO:0000815 ESCRT III complex membership as ok_for_propagation, but the review YAML and notes explicitly reject this ("The PN projection to GO:0000815 should not be added for STX12… it is not an ESCRT-III complex subunit"). STX12 is a syntaxin/SNARE that genetically/functionally intersects ESCRT-III (CHMP2B modifier) but is not a component. GOA has noGO:0000815 for STX12 (confirmed). The other PN mapping (GO:0000045) is fully consistent — review ACCEPTs GO:0000045 + GO:0000407 phagophore assembly site (IMP, PMID:24095276).
PN story / NEW pressure: PN over-reaches on ESCRT-III membership. GO:0000815 is verified real but is a CC for the membrane-scission complex; STX12 is a modulator, not a subunit → projecting membership over-annotates. The autophagy-sealing story is already captured (GO:0000045, already_in_goa). Conclude: PN GO:0000815 over-reaches; do NOT propagate to STX12.
Mapping strategy: The microautophagy "ESCRT-III complex component" leaf should NOT include STX12 (analogous to the TOMM20/RAB7A "broader/wrong-bucket" precedent — here it's a misclassification of an ESCRT modulator as a component). Recommend the PN node exclude STX12 or relabel it as an ESCRT-III modulator (consistent with row-1 context_only).
Evidence alignment: Strong overlap on autophagy: PN's "Syntaxin 13… required for autophagosome maturation" = review PMID:24095276 (core). PN's MDPI "Key Regulators of Autophagosome Closure" review is not in supported_by (secondary). Review enriches with SNARE/endosomal-recycling evidence (UniProt, PMID:19546860).
Verdict: Consistent on the autophagosome-assembly mapping; PN row-3 ESCRT-III-component projection conflicts with the review and is an over-annotation. No gene-YAML edit warranted (review already rejects it correctly); the fix belongs in the PN mapping. Recommended edits: none to STX12-ai-review.yaml; flag PN microautophagy "ESCRT-III complex component" node to exclude STX12 / reclassify as ESCRT-III modulator.