Ghr

PTHR23037 node, shared with the GO:0019955 row. The mechanism the node encodes - ligand engagement by a class-I (haematopoietin) receptor ectodomain coupled to JAK/STAT - is genuinely ancestral and does apply to Ghr, so the transfer itself is sound and the fault is in the term chosen. Checked in the ontology (QuickGO is_a/part_of ancestors): GO:0019221 is NOT among GO:0060396's ancestors, so the proposed replacement is a lateral re-scoping to the ligand-specific pathway rather than a parent-to-child refinement - which is why no GRANULARITY_MISMATCH is asserted here. That ancestry is only partly re-derivable here: the tracked label caches (cache/go/terms.csv, cache/ontologies/go.tsv) carry labels and obsolescence alone, and cache/enums/ carries branch membership rather than is_a/part_of ancestry - GO:0019221 and GO:0060396 both sit in gobiologicalprocessenum, which cannot decide whether either descends from the other. So on this row the QuickGO call really is the whole of the evidence, but for the narrower reason that branch membership is insufficient, not that the repository holds no ontology structure at all. The remaining WITH/FROM entries (MGI, RGD and further UniProtKB accessions, several of which could not be resolved to a protein name here) are by construction further descendants of this same node.

Human interleukin-6 receptor subunit alpha. GO:0019221 is exact for this donor, whose ligand is a bona fide cytokine. Ghr's ligand is the pituitary peptide hormone GH, so the cytokine framing of the pathway term does not carry over even though the JAK/STAT receptor mechanism does.

Human cytokine receptor common subunit gamma, a shared signalling subunit of several interleukin receptors. Same ligand-class argument as IL6R: the generic cytokine pathway term is precise for the donor and imprecise for a hormone receptor.

Human interleukin-12 receptor subunit beta-1. A paralogous class-I cytokine receptor within the same superfamily node; its cytokine-pathway annotation is correct for itself but is the source of the over-broad framing on Ghr.

A different PTHR23037 node from the one behind the GO:0019221 and GO:0019955 rows (PTN002745476); the local PAINT cache carries no tree structure, so their relative depth is not established here and none is claimed. Growth-promoting output is real for Ghr, so the transfer is not false; the objection is that GO:0008284 names a broad downstream physiological outcome shared by most mitogenic receptors and states nothing receptor-specific, which is why the row is marked over-annotated rather than removed.

Human interleukin-6 receptor subunit alpha; a haematopoietic cytokine receptor whose proliferative output is direct and cell-lineage specific. GH's growth-promoting action is largely mediated indirectly through IGF-1 and is tissue- and context-dependent, so the same term is far less informative on Ghr than on this donor.

Human GM-CSF receptor subunit alpha, a directly mitogenic haematopoietic cytokine receptor. Supports the node-level growth-promoting assertion; does not make GO:0008284 a receptor-specific statement about Ghr.

Human interleukin-12 receptor subunit beta-1. Same reading: the broad proliferation term is defensible family biology but is downstream physiology rather than Ghr's evolved receptor function.

The same PTHR23037 superfamily node as the GO:0019221 row. Ligand binding by the class-I receptor ectodomain is ancestral and transfers correctly; what does not transfer is the ligand class. Checked in the ontology (QuickGO ancestors, same answer under is_a alone and under is_a+part_of): GO:0019955 descends from GO:0005515 protein binding and GO:0017046 from GO:0042562 hormone binding, so they are co-descendants of GO:0005488 at different depths and neither is an ancestor of the other. Replacing one with the other is therefore a specificity correction, not a refinement - which is why no GRANULARITY_MISMATCH is asserted here. As on the GO:0019221 row, that ancestry rests entirely on the QuickGO call - here too because cache/enums/ gives branch membership and not ancestry, GO:0019955 and GO:0017046 both sitting in gomolecularactivityenum, rather than because no local file carries GO structure at all. Note the mirror-image case already catalogued in IBA_REVIEW.md pattern 12 Tier A, where IL23R inherits the hormone terms GO:0004925 and GO:0017046 from PRLR across this same over-broad superfamily - the identical node breadth, running in the opposite direction.

Human interleukin-6 receptor subunit alpha. Cytokine binding is exact for this donor; Ghr binds the peptide hormone GH, already captured by the accepted GO:0017046 row on this gene.

Human interleukin-12 receptor subunit beta-1; a paralogous family member whose ligand really is a cytokine.

Human interleukin-23 receptor; resolved from genes/human/IL23R, whose primary accession is this one - a cached gene record rather than the family entries.csv that names the two donors above. The same gene that IBA_REVIEW.md flags as a victim of this node's breadth in the opposite direction; here it is one of the donors whose genuine cytokine-binding annotation leaks the wrong ligand class onto a hormone receptor.

A GHR-clade node in PTHR23037, not the broad class-I cytokine receptor node used by the other Ghr IBA rows. The entire WITH/FROM is GHR itself - UniProtKB:O46600 (bovine GHR) and UniProtKB:P10912 (human GHR) - so there is no paralog, no out-of-family donor and no low-specificity family transfer to argue against; this is an orthologous GHR-to-GHR propagation.

Bovine growth hormone receptor; the IBD seed for GO:0005829 at this node in the local PANTHER PAINT cache. A direct ortholog of the target, so the node placement is a judgement about the GHR clade rather than a similarity transfer from an unrelated protein.

Human growth hormone receptor, the target's direct ortholog and the second WITH/FROM entry.