GMNN source guide

Selected primary Methods, Results and figure descriptions were inspected. Cached-text availability and privately accessible originals are separate facts; normal caches are preserved. No claim is made to have read every cited paper in full.

• PMID:11125146: The author-hosted original and Figure 3–4 were inspected. Human Geminin is full-length; functional licensing assays use Xenopus extracts.
• PMID:16924111: Mouse constructs and endogenous human assays were separated; see Methods and AP4–Gem–SMRT results.
• PMID:19906994: Human fragments and Xenopus extracts support inhibition. Oligomeric switching is a model, not a demonstrated universal endogenous state. The U2OS arrest experiments perturb CDT1.
• PMID:20129055: Flag-GemL26A occupies origins by ChIP. Geminin inhibits the CDT1–HBO1 complex without directly binding HBO1.
• PMID:21543332: Geminin-GFP localization depends on Idas coexpression; this does not determine constitutive endogenous distribution. Idas perturbations are not assigned wholesale to Geminin.
• PMID:22615398: HOXC9 structural and DNA-competition experiments differ from full-length HOXD9 reporter assays.
• PMID:24217620: FUCCI uses a Geminin fragment; the nuclear claim is corroborated independently.
• PMID:41365879: CDT1ΔN/full-length assays differ; steric inference is docking-informed. CDT1 refinements share this evidence.
• PMID:25416956: Human ORF interactions were screened in yeast and retested pairwise. The exact current partner records preserve fragment and isoform labels; study-wide orthogonal validation is not assigned individually to these pairs.
• PMID:31515488: The variant study compares mutant and wild-type interactions in yeast. Its GMNN feature records remain separate from the unsuffixed GO annotation; no disease effect or assayed canonical-sequence equivalence is inferred.
• PMID:32296183: Human ORFs were tested in yeast; the recorded fragments and isoform labels remain distinct.
• PMID:33961781: Original tables corroborate all three incoming pairs in HEK293T, with directed bait/prey roles. These are co-purification associations, not independent demonstrations of direct contact. The replication-complex extension remains literal.

Access limits: PMID:9635433 and PMID:17318181 remain abstract-only in this reading. Positive independently established functions are retained without reconstructing unavailable assays. The exact PMID:23708001 contact remains a question.

Current HPA XML reports supported main nucleoplasm localization and additional cytosol/basal-body locations. Reactome event participants retain their own compartments even when an event-level compartment differs. Leaf-detail requests returned HTTP 403; complete accession-level leaf replay is not claimed.

GO:0140311 describes binding that prevents partner engagement and is the selected mechanistic refinement for CDT1. It descends from GO:0140313, not GO:0140678; no ontology ancestry between those two alternatives is asserted. CDT1 remains a licensing factor, not an enzyme. GO:0140416 describes prevention of transcription-regulator chromatin engagement; AP4-associated corepression remains distinct.

Original-access locators: 11125146 author PDF, Figures 3–4 and Methods; 16924111 Methods/Results; 19906994 Methods/Results; 21543332 Results/Methods; 41365879 Results, Figures 2 and 6–7. These identify the independently inspected originals; they do not change normal-cache availability flags. Eight active cached excerpts are counted at every placement in the source ledger; previous placement counts are historical. Specificity-based TOO_GENERAL judgments preserve the positive experiments.

• PMID:24064211: Human fragments, Xenopus extracts and U2OS assays were distinguished; the proposed oligomeric model is not a universal native stoichiometry.

The six IBA evidence citations specifically refer to paint_ancestry.assertions[*].literal_ibd, with the exact array index and original GAF line stated in each reason. The separately archived raw GOA records are preservation data, not independent biological support.