## CDK5R1
- **UniProt:** Q15078 · **batch:** proteostasis-pr-1217 · **review status:** COMPLETE
- **PN placement:** 1 row, ALP — `Autophagosome closure maturation and lysosome fusion → "Specific function in autophagosome maturation and lysosome fusion unknown"`. **PN-node mapping:** leaf group=no_mapping; class=context_only→GO:0016236; branch=no_mapping. **No GO propagates.** PN Notes: activator of CDK5 required for phosphorylation of ACN (acinus).
- **Consistency:** Consistent. Same "unknown" residual bucket and same Drosophila eLife paper as CDK5, justified only by CDK5R1's role as the CDK5 activator. The review keeps the lone autophagy row (GO:0016241 regulation of macroautophagy, NAS, PMID:21499257) as KEEP_AS_NON_CORE, with the explicit reasoning that the human paper assayed CDK5→SH3GLB1, not a CDK5R1 perturbation — so CDK5R1's autophagy involvement is *inferred* via complex membership, not directly tested. This honest "complex-level context" framing matches PN's no_mapping/unknown placement exactly. No contradiction.
- **PN story / NEW pressure:** Already captured; PN over-reaches no further than CDK5. The defensible-but-indirect GO:0016241 is present (held non-core). No human evidence supports a CDK5R1-specific autophagosome-maturation term, so no ADD. The conserved Acinus hypothesis is routed to a suggested_question/experiment (CDK5R1-specific depletion/rescue in induced-autophagy assays).
- **Mapping strategy:** No change. no_mapping is correct — CDK5R1 is a regulatory subunit whose only autophagy link is via CDK5; propagating GO:0016236 would be doubly indirect. CDK5R1's inclusion in the node adds no new GO-mappable shared function.
- **Evidence alignment:** Divergent in the same way as CDK5. PN cites the Drosophila Cdk5-Acinus-S437 eLife paper; the review's autophagy row rests on human PMID:21499257 (SH3GLB1), which PN does not cite, plus its own notes flagging the inference gap. Convergent on neither autophagy paper directly, but the review is the more cautious.
- **Verdict:** Fully consistent; correctly more conservative than CDK5 (KEEP_AS_NON_CORE vs ACCEPT) given the inference is one step removed. No edits needed.