## UPF2
- **UniProt:** Q9HAU5 · **batch:** proteostasis-batch-2026-06-07c · **review status:** COMPLETE
- **PN placement:** `Translation|Cytosolic translation|Translation termination|Modulation of termination` ; **PN-node mapping:** type `no_mapping`; group `mapped, ok_for_propagation` → GO:0006415 translational termination; class/branch `context_only, too_broad`. Projected: GO:0006415 (goa_status=new_to_goa).
- **Consistency:** Gene-level: consistent. Deep-research/notes, review YAML and GOA agree UPF2 is the NMD adaptor (tandem MIF4G + C-terminal UPF1-binding region) bridging UPF1 to UPF3B/EJC, relieving UPF1 autoinhibition and stimulating its ATPase/helicase. Core annotations: GO:0000184 NMD (multiple, ACCEPT), GO:0035145 EJC, GO:0170010 NMD complex, GO:0003723 RNA binding. But the PN-node frame (translation termination) is not reflected in the review's account of UPF2.
- **PN story / NEW pressure:** PN places UPF2 under "Modulation of termination," projecting GO:0006415 as *new_to_goa*. GOA cross-check confirms UPF2 has NO translational-termination annotation (and no GO:0006449 either). But UPF2 is an NMD adaptor, not a release/termination factor — its termination link is indirect (acts on UPF1 after termination). The story is already captured by GO:0000184 NMD; no NEW translation-termination term is defensible for UPF2.
- **Mapping strategy:** The GROUP→GO:0006415 (translational termination, = polypeptide release) projection is wrong for UPF2 — it would add a peptide-release process the protein does not perform, and would be a genuinely new (unsupported) GOA assertion. This is the clearest over-reach of the UPF set (new_to_goa, no regulatory-term backstop unlike UPF1). The type-level `no_mapping` is correct and should govern; do not project GO:0006415.
- **Evidence alignment:** PN row carries no reference titles; review is well-evidenced for the NMD-adaptor role (UniProt FUNCTION; EJC/complex terms). No competing citations to reconcile.
- **Verdict:** Gene biology consistent, but PN group→GO:0006415 over-reaches and would create an unsupported new GOA term; UPF2's role is NMD adaptor (GO:0000184), not translational termination.

**Recommended edits:** [MAP] do NOT project GO:0006415 (translational termination) onto UPF2 — it is an NMD adaptor recruited downstream of termination, not a peptide-release factor; the type-level `no_mapping` should win. UPF2 is already correctly anchored on GO:0000184 (NMD).