Consistency: Coherent, with TRAPPC4 the strongest autophagy case among the core subunits. Notes, review, PN row ("TRAPP complex, core subunit"), and node mapping agree on core TRAPP membership + RAB1 GEF + ER-to-Golgi transport. Review additionally ACCEPTs GO:0006914 autophagy (IMP, PMID:31794024 patient fibroblasts + yeast Trs23). Neuronal synbindin annotations sensibly split KEEP_AS_NON_CORE (dendrite/synapse) vs MARK_AS_OVER_ANNOTATED (synaptic vesicle, active zone). No contradictions.
PN story / NEW pressure: Unlike the other subunits, TRAPPC4 has gene-specific autophagy evidence (PMID:31794024 "basal autophagy defect and a delay in autophagic flux"), so the existing GO:0006914 autophagy already captures the PN autophagophore-recruitment story. The PN node deliberately does NOT project autophagy (component-bucket scope); the gene's own GOA already carries it. Conclusion: already captured; no NEW term needed.
Mapping strategy: No change. Projected GO:0030008 (OLS-verified) is in GOA exactly. PN-node scope (component only) is appropriately narrower than the gene's autophagy annotation — correctly avoids over-projecting autophagy from the bucket.
Evidence alignment: PN titles ("Membrane Trafficking in Autophagy"; PMID:27066478) both in review. Review adds the disease paper PMID:31794024 (load-bearing for autophagy + RAB1-GEF-essential claims) and the synbindin paper PMID:11018053 — beyond the PN row. No miscitation.
Verdict: Consistent; ACCEPT mapping. Best-evidenced TRAPP subunit; autophagy correctly retained at gene level, not over-projected from the PN bucket.