PN placement:ER proteostasis|Chaperone|HSP70 system|J-domain containing HSP70 cochaperone (branch ER) ; PN-node mapping: type → mapped/ok_for_propagation_to_go GO:0030544 Hsp70 protein binding (goa_status=new_to_goa); all ancestor nodes no_mapping.
Consistency: Consistent on identity, with an MF-term mismatch. Deep research (notes), review and PN agree DNAJC5G/CSP-gamma is a testis-specific, Tdark, completely uncharacterized CSPalpha paralog (J domain aa 17-98), membrane lipid-anchor by similarity, PE2 (transcript-level), PAN-GO 0 annotations. No experimental chaperone data. Mismatch: PN projects GO:0030544 Hsp70 protein binding as the type MF, but the review's core MF is GO:0051082 unfolded protein binding (deliberately cautious family-level call) — GO:0030544 appears nowhere in the review. Same pattern as DNAJC4.
PN story / NEW pressure: PN asserts GO:0030544 (verified real) as new_to_goa — correct that GOA has no chaperone-binding MF (only cytoplasm + membrane IEA). For a J-domain protein this is the canonical co-chaperone MF, defensible by family, but entirely unverified experimentally for CSP-gamma. Conclusion: a defensible family-level ADD candidate, but unverified; the review chose the equally-family-level GO:0051082 instead.
Mapping strategy: Type-level GO:0030544 is appropriate breadth (J-domain co-chaperone), not an over-reach. goa_status=new_to_goa is accurate. The only issue is term-choice harmonization with the sibling paralogs: DNAJC5/DNAJC5B reviews use GO:0030544 as core, whereas DNAJC5G (and DNAJC4) reviews use GO:0051082.
Evidence alignment: PN carries no row references; review cites only the UniProt file (no PMIDs — no functional literature exists). No overlap, no contradiction.
Verdict: Family-level call sound; GO:0030544 is a defensible-but-unverified ADD; harmonize MF term with the CSP family. Recommended edits: [MAP] reconcile the GO:0030544 (PN) vs GO:0051082 (review) MF choice across the CSP paralogs — for consistency with DNAJC5/DNAJC5B, GO:0030544 Hsp70 protein binding is preferable, asserted as predicted/family-level (not experimental).