ATG9B notes

2026-10-05 review (PAINT, affinage)

• Affinage refused to write because of a symbol-collision gate on "yeast". I checked it: a false positive, since the narrative calls ATG9B a functional ortholog of yeast Atg9p and is about human ATG9B. Written with --force.
• PMID:37938170 was a metadata stub in the cache. Re-fetched with full text, which I checked for content ("scramblase" 29 times; Results present).
• Key evidence PMID:37938170; PMID:37938170.
• These calls follow the merged ATG9A review: obsolete PAS membrane → MODIFY to phagophore; piecemeal microautophagy and plasma membrane scrambling → over-annotated; mitophagy and reticulophagy → non-core.
• Bone morphogenesis and programmed necrotic cell death are ISS from mouse Atg9a (Q68FE2), a paralog phenotype, and are over-annotated (they are also over-annotated on ATG9A itself).
• Not used: the many uncached cancer, miRNA and infection papers; the 2026 mitochondrial ATG9B paper (PMID:41811769); and the 2024 neurodevelopmental-variant report.

2026-10-05 revision (reviewer round 1)

• Correction: Q68FE2 is mouse Atg9a (UniProt ATG9A_MOUSE), not rat. I had written "rat" in the review, the notes, the history record and the PR body; it is fixed here, in the review and in the PR body, and the history EDIT records the correction.
• Added propagation_review to all seven ISS rows. The Q68FE2-derived bone morphogenesis and necrosis rows carry failure_modes WRONG_ORTHOLOG_OR_PARALOG and CONTEXT_OR_TISSUE_MISMATCH; piecemeal microautophagy carries LINEAGE_OR_TAXON_MISMATCH.
• Added phagophore (GO:0061908) to the core_functions locations. The autophagosome assembly IDA row has qualifier acts_upstream_of_or_within, while core_functions uses directly_involved_in on the strength of the ATG9A-replacement and scramblase data.
• The ATG9A review's phagophore-membrane NTR is not duplicated here. The autophagy-independent CRC invasion role (PMID:34131310) is uncached and not used.