The genuine isolated normal fetch yielded 50 original assertions, three products and 30 reference identities. All source objects, qualifiers and product order are retained. The displayed human product is P39060-3 (1754 residues), followed by P39060-1 (1519) and P39060-2 (1339). Alternative-sequence reconstruction matches each official FASTA. Their common 183-residue carboxy-terminal endostatin suffix does not establish equivalent expression or activity. The historical amino-terminal NC1-728/NC-493/NC1-303 names differ from the experimentally excised carboxy-terminal NC1 domain. Historic 1516/1336/184-residue descriptions are not silently rewritten as current product boundaries. [PMID:9503365; PMID:14614989]
The single molecular core is the existing extracellular-matrix structural activity conferring tensile strength, in the collagen-XVIII homotrimer at basement membranes. The official term explicitly includes ECM collagens; nonfibrillar collagen XVIII is not thereby excluded. Same-role collagen XV carries the term and network-forming collagen IV has experimental support. Direct human isolated-domain structure and folding demonstrate a trimerization module; the full-NC1 ultracentrifugation assay is mouse. Mouse matrix/retinal experiments connect collagen XVIII to basement-membrane thickness and vitreoretinal anchoring. No isolated whole-human force measurement is claimed. [PMID:19631658; PMID:11927538; PMID:15254016; PMID:21421911]
Human oligomeric NC1 and engineered endostatin dimers promote matrix-dependent motility/morphogenesis, whereas monomeric preparations differ and can antagonize the oligomers. Downstream MAPK/Rac/Cdc42 dependence does not establish that collagen is a receptor, kinase or adaptor. Original inhibitory endostatin preparations, including nonrefolded bacterial material, are positive evidence, while purified native circulating human endostatin lacked the tested antiproliferative effect. These are preparation-specific results, not uniform fragment activities or blanket contradiction. [PMID:11257123; PMID:9008168; PMID:9459295]
The broad angiogenesis/morphogenesis annotations remain non-core. Mouse developmental endostatin assays can show positive proliferation and tube formation, so assigning a universal inhibitory sign would be misleading. No NEW annotation is proposed. A substrate of proteolytic processing is not the enzyme performing that processing, and no precise new molecular activity is inferred from necessity or downstream signaling alone. The complete selected GO-CAM index has no target match; this is not proof that a missing activity should be added. [PMID:15188432; PMID:10626789]
Original human aortic Tables I–III explicitly contain COIA1_HUMAN, with 11/45/6 unique peptides across NaCl/guanidine/SDS fractions. That supports refining the broad extracellular location to ECM. The classification-based tensile annotations are distinguished from what proteomics measured. PMID:20551380
The original stem-cell ECM Table 1 contains COL18A1 at 5.83 normalized spectra in feeder-free human HUES1 ECM, with ND in six feeder columns. The positive target is human HUES1-derived ECM, not a transfer from mouse feeder cells. ND is not treated as universal absence. PMID:23658023
The authentic colon/liver supplement identifies the COL18A1 group in both the complete and matrisome sheets at row 110. The entries repeat one group; accessions include P39060 and legacy/isoform records, so this is not an isoform-resolved measurement or evidence of mechanical force. PMID:25037231
The protected selected caches remain unchanged even when genuine fresh normal full variants were richer. Actual additional Methods/Results reads use their separately archived bytes, with hashes and paragraph locators in the index. The exact COL18A1 targets of PMID:23979707,27559042,28675934 and 27068509 remain unresolved; independent target evidence supports the known ECM/extracellular location. A successful PMID:27068509 supplementary response was a ZIP of 14 images, not the target tables. The PMID:28675934 workbook links returned HTML challenges in the reused authentic attempts. [PMID:23979707; PMID:27559042; PMID:28675934; PMID:27068509]
The complete PMID:22261194 abstract is a porcine model with selected later human validation, not an exclusively porcine study; it does not identify whether COL18A1 was among those human validations. The source-specific link remains unverified while the independently established ECM location is retained. PMID:22261194
The three exosome HDA rows remain UNDECIDED because their exact target/peptide evidence was not recovered. Generic extracellular secretion does not settle vesicle association, and failed retrieval is not evidence of contamination or misattribution. [PMID:19056867; PMID:19199708; PMID:23533145]
The genuine PMID:23962539 publisher PDF Table 1 lists endostatin, explicitly the collagen-XVIII C-terminal product, with vascular elastic fibres. The owner inspected extracted target lines; the independent reviewer also viewed the actual page. This supports a non-core fragment association, not every intact precursor or an elastic-fibre assembly function. The protected normal cache stays abstract-only. PMID:23962539
The skeletal IBA has real positive collagen XXVII descendant experiments, including zebrafish notochord/vertebral effects. The historical PAINT node and a target-specific COL18A1 skeletal step remain unresolved; the annotation stays UNDECIDED without a wrong-paralog, donor-count or evidence-of-absence argument. Current literal source identities and records are retained in the structured index. PTN source resolution is separate from confidence in other target functions. PMID:20041163
Rat ticlopidine-associated serum endostatin has functional anti-endostatin neutralization evidence in HUVEC assays. This supports a contextual xenobiotic response, not metabolism or a drug sensor. Rat tendon-cell pressure experiments report increased endostatin and inhibitory conditioned medium, but the available abstract does not establish that endostatin uniquely accounts for the medium effect. Both response rows remain non-core with species and assay scope explicit. [PMID:11353854; PMID:12798059]
Current ARBA conditions match morphology through three FunFam entries, proliferation regulation through those entries plus the non-fungal condition, and ER lumen through two InterPro entries plus primates. The initial parser used the wrong database name for FunFam; its output is archived and explicitly superseded by the corrected literal comparison before curation. An unmatched current elastic-fibre rule or an unavailable current rule does not reconstruct historical failure. The actual endostatin review row independently supports the limited elastic-fibre location.
Eight actual Reactome events and nested target reference paths support precursor or fragment locations. ER residence is biosynthesis, and secretion, cleavage and degradation events do not confer the processing enzymes’ activities on the collagen chain. One current P39060 fragment has endCoordinate 11754 despite a 1754-residue displayed sequence; this occurs in the release and laminin-dimer paths and is retained as a model limitation, never used for sequence mapping. Human circulating-fragment evidence retains broad extracellular-region scope rather than being forced into an exclusive matrix location. PMID:10441114
The biological decisions remain unchanged. Eleven short, nonduplicated primary anchors now identify the key source passages in the annotation reviews, and all 16 HIGH-relevance references have scoped findings. The human aortic and HUES1 numerical measurements are supported by their original target tables; they remain localization measurements, not mechanical assays. Sources that were available only as abstracts or separately recovered originals retain their recorded access boundaries.
The construct distinction is explicit in the original oligomer study: PMID:11257123. This statement concerns the tested motility assay, not every activity of monomeric endostatin. Human isolated-domain trimerization is separately delimited by PMID:19631658. Neither experiment demonstrates whole-precursor force generation.
The chick basement-membrane study is added as supporting ortholog evidence, with its species stated. Knobloch occipital encephalocele is explicitly considered for the skeletal IBA; the available clinical abstract does not resolve the specific skeletal contribution or the historical PAINT placement, so UNDECIDED is retained without a biological exclusion claim. Six electronic-rule reasons now point to the published index and its literal rule/target-feature comparisons.
Repeated product descriptions are retained because they document exact current-sequence boundaries and shared fragment limits; no isoform-specific activity is inferred from historical names or tissue expression.