## VPS37B
- **UniProt:** Q9H9H4 · **batch:** proteostasis-pr-1217 · **review status:** COMPLETE
- **PN placement:** 1 row, ALP — `Autophagosome closure maturation and lysosome fusion → Sealing of autophagophore membrane → ESCRT-I complex component`. **PN-node mapping:** leaf=mapped→GO:0000813 ESCRT I complex; sealing group=mapped→GO:0000045 autophagosome assembly; class=context_only (GO:0016236). **Projected:** GO:0000045 (more_specific_than_existing_goa), GO:0000813 (already_in_goa_exact).
- **Consistency:** PARTIAL CONTRADICTION. ESCRT-I membership is fully consistent (review ACCEPTs GO:0000813 across many evidence lines; deep notes confirm TSG101/VPS28 co-IP, PMID:15218037; structural headpiece PMID:32424346). But on autophagy the review and PN diverge: the review MARK_AS_OVER_ANNOTATED the GO:0016236 macroautophagy rows and does NOT add GO:0000045, explicitly stating the phagophore-closure role is VPS37A-specific and should not be transferred to VPS37B. PN nonetheless projects GO:0000045 to VPS37B via the shared sealing group.
- **PN story / NEW pressure:** Over-reaches for autophagy. PN's group-level GO:0000045 projection asserts an autophagosome-assembly role the review judges unsupported for VPS37B (no gene-specific phagophore-closure data; only "probably autophagy" review-level statements). ESCRT-I (GO:0000813) is correctly already captured. No defensible NEW term beyond GO:0000813.
- **Mapping strategy:** Flag. The leaf→GO:0000813 mapping is right and harmless (already in GOA). The group→GO:0000045 projection is the issue: it is broader than the review supports *for this paralog*, paralleling the rejected TOMM20/HSPA8/RAB7A "shared-node over-reach" precedent. GO:0000045 should not propagate to VPS37B on current evidence — restrict the sealing-group projection or mark VPS37B's group membership as non-propagating.
- **Evidence alignment:** Partial. PN cites only PMID:32424346 (ESCRT-I scaffold), which the review uses for complex membership but explicitly notes used a VPS37B headpiece for general mechanism, not VPS37B autophagy. Review adds PMID:15218037, 23924735 (ALG-2) not in PN.
- **Verdict:** Consistent on ESCRT-I; GO:0000045 projection over-reaches. **Recommended edits:** none to YAML (review correctly omits GO:0000045); flag PN sealing-group→GO:0000045 as non-propagating to VPS37B.