## ATP23
- **UniProt:** Q9Y6H3 · **batch:** proteostasis-batch-2026-06-03 · **review status:** COMPLETE
- **PN placement:** `Mitochondrial proteostasis|Organelle-specific protein degradation|Intermembrane space protease` (group = mapped, GO:0005758 mitochondrial intermembrane space, more_specific_than_existing_goa); parent *class* `...|Organelle-specific protein degradation` = mapped, GO:0035694 mitochondrial protein catabolic process (new_to_goa).
- **Consistency:** Deep research (falcon), review, and PN agree on the core model: conserved M76 metalloprotease / ATP synthase Fo assembly factor and mitochondrial protein-processing peptidase; KUB3/XRCC6BP1 DNA-repair annotations are alias-driven and REMOVEd. One **divergence**: PN class projects GO:0035694 *catabolic* process, but the review deliberately uses GO:0034982 mitochondrial protein **processing** (maturation, not catabolism) and never adds GO:0035694.
- **PN story / NEW pressure:** PN adds GO:0005758 (IMS) as a NEW location — the review independently adopts this (action: NEW, ISO, conservative topology inference). Verified real and appropriate. No over-reach at the leaf. **ADD (IMS) — already implemented in review.**
- **Mapping strategy:** The leaf (IMS protease → GO:0005758) is sound and gene-appropriate. The **class-level GO:0035694 over-reaches**: its OLS definition is "breakdown of a mitochondrial protein via an intramitochondrial lysosome-like organelle to eliminate damaged/oxidised proteins" — ATP23 is a *processing/assembly* peptidase, not a catabolic degrader. This is the TOMM20/HSPA8 "broader/wrong-flavor parent" failure mode. Recommend the class projection not propagate to ATP23.
- **Evidence alignment:** PN gives no reference titles; review anchors on PMID:17135290 + PMID:17135288 (yeast Atp23 processing + Fo assembly), PMID:34800366 (MitoCoP). GO:0035694 and GO:0005758 are both absent from goa.tsv (new), consistent with dossier.
- **Verdict:** MOSTLY CONSISTENT — IMS location good; class-level *catabolic* projection mis-flavored for an assembly/processing peptidase. **Recommended edits:** [MAP] mark GO:0035694 class projection as not_for_propagation to ATP23 (gene is processing/assembly, not catabolism); review's GO:0034982 is the correct process term.