This audit preserves the five published machine-source assertions, both original publication caches, genuine Falcon report and complete OpenScientist hypothesis subtree. There was no published notes file to append to. The starting review at main 3fbffb26407414459e21ed806ebfa28885e8d809 contained five source assertions and three reviewer-authored NEW processes. Its existing COMPLETE flag does not substitute for this evidence audit. The original source family and the newly fetched comparison are distinguished in the source evidence record.
The current official human identity is CFAP300, HGNC:28188, UniProt Q9BRQ4; C11orf70 is the previous symbol and DNAAF17, FBB5 and MGC13040 are aliases. The new normal fetch returned six source assertions: the old single cytoplasm ISS assertion now separates the Paramecium A0CY51 and Chlamydomonas A8IYS6 donors, while the fresh rows also provide qualifier/WITH fields absent from the old YAML. All five historical source objects remain exact. The new projection is recorded for future source maintenance rather than silently replacing the original evidence history.
The normal fetch also supplied three authentic products, Q9BRQ4-1, -2 and -3. They are already represented in the protected UniProt record, but the historical YAML has no product list. That absence is retained in this bounded audit; it must not be described as equality with the fresh three-product intake. No tested isoform or co-IP host is inferred from the current canonical sequence.
PMID:29727692. The genuine UCL accepted manuscript was read at the localization/fractionation, RNAi and IFT Methods/Results and Figures 3–5 descriptions. Patient nasal-cell TEM and dynein-marker imaging show combined outer/inner-arm defects. The model experiments express each species' own tagged ortholog: Paramecium C-terminal GFP and Chlamydomonas FBB5 with a C-terminal triple-HA tag. Approximately 98% is cytoplasmic, with a small ciliary pool released into the detergent-soluble membrane/matrix fraction. IFT perturbation and regeneration experiments establish the distribution of CFAP300 itself; the paper calls that distribution [PMID:29727692, "IFT dependent"]. They do not show a CFAP300 molecule carrying a dynein complex. IFT46 N-terminal deletion reduces ODA16/ODA abundance without reducing FBB5. The ortholog data are not direct human localization experiments.
PMID:29727693. The actual publisher article-plus-supplement and Figure S12/Table S2 were inspected. A DNAAF2 yeast-two-hybrid signal is corroborated by recovery of DNAAF2-3xFlag with C11ORF70-HA. PIH1D3 and IFT46 yeast candidates were excluded because of autoactivation and were not recovered in the co-IP. The selected assay descriptions state sequence confirmation against RefSeq but do not identify the exact accession/version or co-IP host. These results support an accessory role in assembly machinery while leaving the performed molecular step unresolved. The authors themselves describe a [PMID:29727693, "preassembly factor"] interpretation and also leave a possible delivery role open. The preserved GOA IPI cites the neighboring PMID:29727692; its precise historical linkage remains unresolved. Keeping the scientifically supported binding assertion uses independent PMID:29727693 evidence without changing that original field.
PMID:36246608. Full Methods, Results and Discussion, and the actual Figure 4 image, were read. Endogenous antibody staining in normal human nasal epithelial cells supports a predominantly cytoplasmic distribution and puncta along motile cilia, with substantially depleted staining in patient cells. The antibody epitope spans residues 113–188; the paper discusses possible residual signal from a shorter transcript. Airyscan images are endpoint localization data, not trajectories or a dynein-cargo binding assay. The Discussion describes a [PMID:36246608, "potential role in transport"]. Basal-foot disorientation could be secondary to immotility, so it does not establish a new polarity function.
PMID:30916986. The exact-main cache remains abstract-only, but author-uploaded indexed full Methods, Results and Discussion were accessible. Patient ODA/IDA markers accumulate abnormally near the apical cell region, with DNALI1 and DNAH5 differing relative to centrin. This motivates a trafficking model but does not assay whether the retained material is a preassembled complex. Planarian RNAi addresses an ortholog, not human CFAP300. The supplemental methods and figure pixels were not independently recovered. No transport-process proposal is based on these endpoints.
Both cytoplasm assertions, the axoneme assertion and the motile-cilium assertion remain ACCEPT, supported by actual localization rather than disease necessity alone. The axoneme IEA source is curator-backed and human Airyscan microscopy detects a small CFAP300 pool along the ciliary axoneme; this minor ciliary pool does not make axonemal localization the protein's core activity, but it is enough to keep the existing cellular-component assertion without duplicating the accepted motile-cilium row.
The generic binding assertion remains KEEP_AS_NON_CORE under the pinned ClinGen instruction. DNAAF2 association is positive, but no more specific chaperone, motor, dynein-binding or cargo-adaptor MF is established. The current IntAct whole-target query returned six unrelated-source records and no exact 2018 pair; this does not negate the primary Figure S12 experiment.
Retain the two prior authored assembly proposals, GO:0036158 and GO:0036159. CFAP300's inferred participation as an accessory assembly factor rests on the combination of selective conserved arm defects and association with DNAAF2, not on necessity alone or on a demand for purified catalysis. The precise preassembly, stabilization, delivery or incorporation step remains open. Live same-role comparator queries show human DNAAF1, DNAAF2 and DNAAF3 carry dynein-assembly annotations, with DNAAF1 carrying both specific child processes. These comparators support the applicability of assembly processes to accessory factors; they do not prove an identical mechanism. The fixed-main GO-CAM index has no CFAP300/accession match. Current typed parents show that both child terms descend from GO:0070286. Withdraw only that redundant prior authored parent proposal, preserving its full old object in the audit checkpoint and its immutable source URL/hash in the evidence record. This is not a claim that the broader biological description is false.
The assembly-centred core deliberately leaves the molecular activity unresolved. The structured knowledge gap makes the same distinction. The proposed MF term is therefore mechanism-neutral: it requests an activity term for DNAAF-like cytoplasmic preassembly factors without asserting that CFAP300 is specifically a scaffold, co-chaperone or transport adaptor.
The genuine historical Falcon and OpenScientist files are byte-preserved. Their broad cargo-delivery language and the OpenScientist claim of an IFT139 physical interaction are not treated as primary proof; the inspected experiment perturbed IFT139 and measured localization. No artifact of those providers has been manually edited. A genuine isolated Falcon attempt timed out and the explicit fallback failed with an insufficient-quota response. No replacement provider report was authored. This manual audit follows the successful primary recoveries and preserves failed-access receipts as provenance.
The two original publication caches and the two additionally selected existing-main caches are unchanged. External primary access is disclosed separately from cache completeness. Short exact anchors are consolidated across YAML and this new notes file; the quote audit counts every repeated authored occurrence. The source evidence record contains the official donor records, assay locators, typed term checks, hashes and original-versus-fresh projection rather than uncheckable operational dependencies.
This section reconciles external follow-up commit 82f59c49ada72e0299ddb5fe9445ba95d5a99973 and preserves its entire published notes prefix exactly. That commit already corrected the earlier axoneme downgrade; this section retains the correction and clarifies the remaining evidence, molecular-function and access questions. The original five machine objects, two retained authored assembly proposals, all nine reference identities, absent historical product field, source caches and genuine provider artifacts remain unchanged. All seven current actions are preserved, including the external commit’s ACCEPT for axoneme. Its restored findings are retained as informative, source-scoped summaries rather than repeated long excerpts.
Axoneme evidence. The current GO:0005930 definition includes associated proteins, so stable incorporation as a structural subunit was an unnecessarily strong acceptance condition. The full human primary study PMID:36246608, Figure 4 and Immunofluorescence Methods uses endogenous antibody staining, alpha-tubulin colocalization and Airyscan microscopy in control and patient nasal cells. It supports localization along the axoneme and independently corroborates the UniProt location. The assay does not resolve direct microtubule contact, residence time, structural stoichiometry or dynamic dynein-cargo transport. The soluble ciliary pool in the distinct tagged-ortholog fractionation of PMID:29727692 limits structural interpretation but does not override the human localization. Accepting the original row also removes the unnecessary replacement with the already accepted motile-cilium term; duplicate evidence for a valid location is not itself forbidden.
Findings and quotation scope. Informative findings are retained and expanded for both 2018 primary papers, with additional findings for the human 2022 and 2019 studies. They distinguish actual measurements from inferred mechanism and retain access/construct limits. Short exact anchors are allocated to the axoneme decision, assembly proposals, DNAAF2 association, core and selected reference findings. A source may be cited elsewhere without repeating the same passage. The aggregate 25-word limit per primary source remains binding for newly authored excerpts, including repeated occurrences; it is not an instruction to discard scientific findings. The audit separately records introduced YAML/appended-note excerpts and the unchanged legacy-note prefix, and also checks the stricter combined total. No long historical provider text is copied into new evidence.
Biological uncertainty versus an ontology gap. The assembly-process participation assessment is unchanged. CFAP300 has a DNAAF2-associated accessory role, but the performed molecular step remains unresolved. This is a biological MF_DARK gap: a curator cannot close it by merely naming a new activity. The exact-head CLAUDE rule for proposed_molecular_function applies when a supported activity lacks a suitable term; it does not require converting every unknown activity into an ontology proposal. Direct folding assistance, selective bridging of two clients, and a defined preassembly-specific molecular step have not been established by the inspected experiments. Therefore neither a chaperone/scaffold claim nor a renamed assembly process is proposed as an MF. The earlier wording that rejected the old term solely for its illustrative scaffold/co-chaperone examples was too narrow: the decisive issue is that the molecular activity and preassembly-versus-delivery stage remain unknown. The intervening commit’s proposed preassembly-factor term and matching core slot are therefore withdrawn: they asserted a missing activity and a stage that the same evidence leaves unresolved. This does not negate the accepted accessory role in assembly. An eventual demonstrated activity may justify an existing term or a precise NTR.
The scientific questions now concern product-specific function and possible DynAP participation; the historical PMID:29727692 IPI linkage and exact old clone identity remain curation provenance issues documented above. The restored DynAP experiment tests enrichment, residence dynamics and client assembly with interaction mutants; it is a proposed test, not a claim that CFAP300 was already shown in these structures.
Access metadata. full_text_unavailable describes access to the paper, not whether a repository cache contains its complete text. It is restored to false for the externally accessed 2018/2019 primaries, superseding the intervening commit’s cache-based flags: the genuine article/supplement or indexed author text was inspected, as documented in the source evidence. The normal caches stay byte-for-byte unchanged and their own full_text_available metadata is not reinterpreted or edited. The 2022 cache itself contains full Methods/Results. The earlier reference access notes remain applicable.