## UMAD1
- **UniProt:** C9J7I0 · **batch:** proteostasis-pr-1217 · **review status:** COMPLETE
- **PN placement:** ALP `…|Sealing of autophagophore membrane|ESCRT-I complex component` ; **PN-node mapping:** leaf mapped/ok_for_propagation GO:0000813 ESCRT I complex; group mapped GO:0000045 autophagosome assembly (both flagged new_to_goa).
- **Consistency:** MAJOR divergence. The PN annotation asserts UMAD1 is "Component of the ESCRT-I complex, involved in autophagosome closure" and projects GO:0000813 + GO:0000045. The review YAML and notes conclude the opposite: UMAD1 is poorly characterized, UniProt has no FUNCTION comment and zero GO/PAN-GO annotations, and the only GOA evidence is generic IPI protein binding (HuRI; GABARAPL1, TH isoform 3). The review explicitly declines any ESCRT-I/autophagy function.
- **PN story / NEW pressure:** The PN asserts a role (ESCRT-I membership + autophagosome assembly) not in GO and not supported by the review. PMID:32424346 lists UMAD1 only as a *possible* UMA-domain fourth subunit and notes some theoretical complexes may not form. Verdict: PN over-reaches — neither GO:0000813 nor GO:0000045 is defensible for UMAD1 on current evidence (would be a speculative NEW from name/domain only).
- **Mapping strategy:** This gene should NOT drive the ALP leaf/group projection. UMAD1 is the weakest member of the ESCRT-I-component leaf (cf. UBAP1, VPS28); propagating GO:0000813/GO:0000045 to it on family-name grounds is exactly the kind of paralog/domain over-annotation to avoid.
- **Evidence alignment:** PN cites PMID:32424346 (helical ESCRT-I scaffold). Review cites the same PMID plus PMID:32296183 (HuRI) and UniProt — and reads PMID:32424346 as explicitly NOT establishing a UMAD1 complex. Same papers, opposite conclusions.
- **Verdict:** CONTRADICTION — PN projects ESCRT-I + autophagosome-assembly membership that the review rejects as unsupported. Recommend the PN leaf not project GO terms onto UMAD1 (mark gene-level no_mapping/UNDECIDED) pending direct evidence.