Consistency: PARTIAL TENSION. Deep research and review YAML converge on RBJ as a chimeric Ras/Rab-GTPase + J-domain protein whose CHARACTERIZED function is a GTP-bound MEK/ERK scaffold/positive regulator (core MF GO:0005525 GTP binding; core BP GO:0070374 ERK cascade), oncogenic in GI cancer. The C-terminal J-domain is only PREDICTED to recruit Hsc70/HSP70 (PMID:14980719), with no experimental HSP70-cochaperone characterization. The PN node files RBJ purely as a "J-domain HSP70 cochaperone," which captures the J-domain but omits its dominant, evidence-backed GTPase/ERK identity — a placement/emphasis mismatch, not a hard contradiction.
PN story / NEW pressure: PN asserts GO:0030544 Hsp70 protein binding (verified real). GOA lacks any HSP70/chaperone-binding term, so the "more_specific_than_existing_goa" label is loose — there is nothing in GOA for it to be more specific than; functionally it is new_to_goa. GO:0030544 is defensible only as a domain-level prediction; over-reaches if presented as a core function, since the gene's real MF is GTP binding / ERK scaffolding.
Mapping strategy: Node status unchanged. goa_status flag ("more_specific_than_existing_goa") should be new_to_goa. The projected term is domain-inferred, not characterized — keep IBA/IEA strength, do not elevate to core.
Evidence alignment: PN row carries no titles; review cites PMID:14980719 (RJL family, J-domain prediction) and PMID:24746703 (nuclear MEK/ERK entrapment) — both VERIFIED. The HSP70-binding story rests only on the family prediction; no shared experimental paper supports it.
Verdict: Consistent core review; PN placement over-emphasizes the J-domain. GO:0030544 is a defensible inferred ADD but not core. Recommended edits: [MAP] change goa_status from more_specific_than_existing_goa to new_to_goa (no existing chaperone-binding GOA term to refine).