Manual assessment, 2026-10-10. This is an evidence locator, not a provider-generated report. Zero primary-source quotations are introduced here. Source rows use zero-based indices in the original 17-row seed.
| Source | Access and exact relevant evidence |
|---|---|
| PMID:21399614(https://doi.org/10.1038/emboj.2011.63) | Exact existing-main full cache: selected human centrosome proteomics/localization Results and KIAA0562 discussion. Original target figure/table pixels were not newly inspected. Independent human localization is resolved by the following paper. |
| PMID:23970417(https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3820246/?report=xml) | Authentic full HTML; selected human Results, Figures 5–8 descriptions, supplementary S1 description and RNAi Methods. RPE1 distal centrioles/ciliary tips, human tracheal ciliary tips, and mitotic spindle poles are distinguished from Chlamydomonas ultrastructure. Download SHA256 47fb80f681db1d081913840dd133257a49e2aa67786796848655f7cdf637cc8a. Normal cache remains abstract-only. |
| PMID:26420826(https://doi.org/10.1093/nar/gkv952) | Exact existing-main full cache: Y2H Methods and explicit KIAA0562/CEP104 discussion. Human SF1 residues 1–441 was the bait; the positive human CEP104 fragment records and yeast host are literally extracted in the adjacent JSON. This is not an endogenous splicing experiment on CEP104. |
| PMID:26638075(https://doi.org/10.1016/j.cell.2015.10.065) | Normal abstract plus exact human BioID records: CEP104 bait and CCDC66, CSPP1 or TOGARAM1 prey in HEK293 cells. This is proximity evidence. Later targeted co-IP provides independent association support; the BioID records are not relabeled as those later experiments. Full article and supplement were not newly read. |
| PMID:32453716(https://www.jci.org/articles/view/131656) | Publisher full HTML plus authentic supplement. Selected Results, supplement Methods pp.5/7 and Figure S1 image pp.15–16 inspected. Human CEP104 construct is NM_014704.4. Reciprocal Flag/HA co-IP in HEK293T with mRFP controls tests TOGARAM1 (A/D), CCDC66 (B/E), CSPP1 (C/F), ARMC9 (G and main Fig.1F), and CEP104 itself (H/I). Supplement SHA256 39d18d72abfdce7d327d5275f7d614e8e67e248f05ece1a8c4bc21819bf322fa. This is cellular association, not purified binary binding or fixed stoichiometry. |
| PMID:36606424(https://doi.org/10.1242/jcs.260327) | Final published abstract inspected; final body access failed. The separately inspected September 2022 primary preprint, DOI10.1101/2022.04.10.487777, Results pp.13–15/Figure5/S4, describes controlled GFP-CCDC66 co-IP and mNG-CEP104 in RPE1 cilia. It is not represented as the final article. The association and localization also have independent full-primary corroboration. |
| PMID:27402853(https://pmc.ncbi.nlm.nih.gov/articles/PMC5000094/) | Normal abstract plus externally retrieved full selected Results/Figure2/Methods: human TOG418–673 binds bovine tubulin; the solved TOG428–686 structure is chicken. No human tubulin-bound crystal structure is claimed. |
| PMID:28017521(https://pmc.ncbi.nlm.nih.gov/articles/PMC5222566/) | Genuine full normal cache: selected Figure1 Results and construct/SEC-MALS Methods distinguish human367–685 biochemical TOG from human392–676 crystal construct. Interface mutations reduce tubulin binding; NEK1 association does not confer kinase activity. |
| PMID:32820051(https://pmc.ncbi.nlm.nih.gov/articles/PMC7586207/) | Normal abstract plus external full selected Results/Methods and Figure2 image: isolated TOG400–690 and porcine-tubulin bulk turbidity, separate full-length925 rescue in RPE1 cells. Inspected Methods do not explicitly identify the construct species/accession. The bulk assay does not measure full-length single-filament elongation kinetics. |
| PMID:39856351(https://doi.org/10.1038/s41594-025-01483-y) | Exact existing-main full cache. Full-length human925 CEP104, tagged/purified from HEK293T, was assayed on porcine-tubulin microtubules. Selected Results/Methods and Figures1–3 establish plus-end blocking and inhibition of shortening, enhanced by recruitment partners. The combined module contains mouse TOGARAM1. Extended Data2c supports CCDC66/CSPP1 association; Extended Data4c does not reproduce the earlier weak ARMC9 association. Independent activity peer inspected Figure3 directly. |
For the ARMC9 assertion (row7), both original positive evidence and later non-replication are retained; the outcome is UNDECIDED. No erroneous-curator inference follows from an empty database query or the later result alone. CEP104 self-co-IP is retained at assay scope without a native oligomer claim.
The sole NEW term is GO:0140778, supported by directly measured stabilizing activity. Its current is_a/part_of ancestors are GO:0140777 and GO:0003674, not the existing protein-binding terms. A separate NEW plus-end-binding term would be redundant with an existing ancestor and is not added. No NEW biological process or stable-complex annotation is proposed. The pinned main GO-CAM index contained no O60308/CEP104 entry.
The earlier no-BP decision is superseded by the mapped process evidence described in the appended notes. Full external PMID:32820051(https://pmc.ncbi.nlm.nih.gov/articles/PMC7586207/) Figure 2D–F, Results and cloning/rescue Methods were reread: siRNA-resistant full-length YFP rescue is compared with W448A/V493D/R626A; the Methods do not state the cDNA species/accession. The reported ciliary localization of both constructs is distinct from the loss of rescue. Human endogenous CEP104 depletion, the isolated TOG400–690/porcine-tubulin turbidity assay and defined human biochemical proteins in independent studies are not conflated. The process inference is regulation of elongation/assembly extent, supported by CEP104's own tubulin interaction and end regulation, not by a substrate or generic necessity argument.
Current term definitions were retrieved on 2026-10-10. GO:0097542 is ciliary tip; GO:0015631 includes soluble and polymeric tubulin binding. Binding ancestry includes GO:0005515 for both proposed binding alternatives. Consequently the explicit NEW ancestor/descendant rule precludes an additional binding assertion, while a separately supported core MF is appropriate. None of the partner-specific source rows actually tests a tubulin partner, so none is relabeled to manufacture that refinement.
Process ancestry and the bounded five-protein comparator query were also read. The response had 11 hits, one complete page: CCDC66 4, TOGARAM1 3, ARMC9 4, CEP104 0, CSPP1 0. The experimental CCDC66 and TOGARAM1 assembly assertions provide comparable ciliary microtubule-regulator roles; this is not a global all-species absence survey or a claim that every comparator carries GO:1902017. Response SHA256 c7b53a467eee4e896d481626aeae2df2c6b66542d33d24f90a562b52d5790f7a. The exact cb9dd513246e54396fcd4fc2c2ebaaac546297bb GO-CAM index had no O60308/CEP104 match. No already-carried or proposed ancestor/descendant BP duplicates the selected GO:1902017.
The raw GOA remains byte-identical, including both IntAct and UniProt records for PMID:32453716/TOGARAM1. Only ASSIGNED BY and DATE differ between TSV lines 14 and 18; normal-source YAML retains one corresponding assertion. All 17 original source assertions and three product objects are protected, as are the ten authentic publication caches. The original interaction JSON retains its source URLs and exact extracted fields. The source-specific ARMC9 uncertainty is now also represented as a DISPUTED finding under PMID:32453716, supported by the later context-limited non-replication; all original actions remain unchanged.