This review preserves all 92 machine-derived assertions, all three alternative products and all 52 original reference identity pairs. The immutable genuine seed is e2df0827119ff0b400914f138097480ab6ff2eed250422648d04ac5c37885fe1. Current decisions are {'KEEP_AS_NON_CORE': 25, 'UNDECIDED': 7, 'MODIFY': 8, 'ACCEPT': 52}; there are no NEW annotations and one integrated structural core. Existing source rows are not collapsed, including repeated terms with distinct evidence.
The isolated normal fetch completed successfully. The default Falcon attempt and configured perplexity-lite fallback both terminated during dependency-download/DNS failures; no provider report was fabricated. The genuine command, output and failure receipts are retained. Manual primary review supplies the scientific assessment. Source selection preserves exact applicable main bytes over fresh variants; authentic fresh full variants remain archived for reading where a protected main cache is abstract-only. Unused incidental PMID:15489334 is not selected for publication.
The target is HGNC:2200 / UniProt P02458. The displayed product P02458-2 is 1,487 residues; the independently reconstructed P02458-1 is 1,418 residues and P02458-3 is 268 residues. All three original product objects remain intact. Sequence equivalence of the screened C-terminal segment and current product 3 does not prove that natural product 3 was the tested source: the original screen assertion explicitly identifies P02458-1. No gene-wide location or assembly conclusion is silently applied to every alternative product.
The single core uses the existing collagen tensile-strength term GO:0030020, ECM location and type-II trimer context, with the existing fibril-organization process. COL2A1 also supplies the alpha-3(XI) chain. Official collagen guidance and actual matrix incorporation support the structural class; mass-spectrometry detection is not a direct tensile measurement. The original human fetal-cartilage study PMID:8660302(https://pubmed.ncbi.nlm.nih.gov/8660302/) includes directly sequenced alpha-1(II) telopeptides and II–IX cross-links despite a title foregrounding type IX. It does not settle the specific homodimerization assertion, which remains UNDECIDED; positive homotrimer assembly and broader identical-chain association remain intact.
The mouse IIA study PMID:24113490(https://pubmed.ncbi.nlm.nih.gov/24113490/) establishes type-II and XI incorporation at its actual engineered mouse splice-product scope. PMID:30464261(https://pubmed.ncbi.nlm.nih.gov/30464261/) uses collagen peptides with GPP stabilization and T4 foldon, tests trimer formation and COMP/TSP-4 binding, and does not reconstruct complete native human fibrillogenesis or solve a collagen-COMP complex. The independent ROOT core consultation is explicitly bounded to its actual source selections; no original figure-pixel read is claimed.
The original PMID:7829510(https://pubmed.ncbi.nlm.nih.gov/7829510/) human achondrogenesis abstract supports altered cartilage matrix but does not settle the precise chondrocyte-differentiation step without full experimental access. A production GO-CAM gives collagen structural activity a part-of relationship to differentiation using that same source. This is a considered curator assertion, not an independent experiment, and is preserved while the exact source-specific process remains UNDECIDED. Mouse cartilage can differentiate and hypertrophy without collagen II despite abnormal architecture; that does not disprove a contextual human matrix contribution.
Independent PMID:9832566(https://pubmed.ncbi.nlm.nih.gov/9832566/) full Methods/Results establish persistent notochord, failed disc formation and altered vertebral matrix in Col2a1-null mice after initially normal early development. This supports retention of an existing contextual notochord role. The proposed mechanical/swelling-pressure explanation was not directly measured. Original mouse expression donors for CNS development, otic-vesicle development and limb-bud formation do not resolve the precise work performed; those source assertions remain UNDECIDED. Human skeletal, visual, hearing and cartilage phenotypes are retained as context where supported; no phototransduction requirement is imposed on a structural ocular protein. The exact embryonic joint-morphogenesis source remains unresolved.
Eleven original screen assertions join 33 literal 42-field records. All contain positive array, pooling and validated-Y2H records with human participants and yeast host. COL2A1 is the recorded P02458-1 C-terminal 1151–1418 segment. PRPF40A-2 is additionally restricted to 716–930. Four class refinements describe the bound product: gelsolin as an actin-regulatory protein, FGFR3 as a receptor tyrosine kinase, CASP6 as a caspase-family precursor and GRIN2C as an ionotropic glutamate-receptor subunit. They do not establish mature processing, domain-specific contact, partner activation or a native collagen signaling pathway. Canonical gelsolin is the secreted/plasma precursor; no cytoplasmic-isoform substitution is made.
The remaining seven generic screen interactions are retained as non-core under the campaign policy in projects/CLINGEN_MENDELIAN.md. Genericity and expected compartment separation do not invalidate positive experiments. HIP1 has supported F-actin association, but the reviewed evidence does not establish the cytoskeletal-regulatory class; the PRPF40A interaction remains scoped to its recorded C-terminal fragment; LAMP2B is not LAMP2A; KEAP1 is not an isolated E3 enzyme; UBQLN1 degradation-shuttle identity does not by itself establish a protein-folding chaperone class. Q9Y649 remains an exact short reference product without an invented gene/function. For COQ8A, actual primary and curated evidence distinguish atypical kinase-like structure from settled wild-type human trans-protein-kinase chemistry. Positive binding is retained without forcing that subclass.
The genuine PMID:32296183(https://pubmed.ncbi.nlm.nih.gov/32296183/) HuRI selections and the separate PMID:32814053(https://pubmed.ncbi.nlm.nih.gov/32814053/) neurodegenerative-disease screen are not conflated. The latter's protected cache is abstract-only; original author-manuscript STAR Methods were read from the authenticated MDC PDF, page e8. Global mammalian validation performance is not assigned to each pair. No independent binding consultation is claimed; BrCa was assigned another whole review before opening the fixed input.
The HLA-DR4 source binds a processed human collagen-II peptide, not a native fibril. Lubricin assays distinguish human cartilage proximity from direct bovine-collagen-II/human-recombinant-lubricin binding. PDGF binding remains positive at complete-abstract scope with original reagent details unresolved.
PMID:20603131(https://pubmed.ncbi.nlm.nih.gov/20603131/) was read across its complete selected experimental body. Its constructs, primers, antibodies and measured chains are type-I COL1A1/COL1A2; no COL2A1 assay was found. The original COL2A1 source object is retained and the reference is marked MISCITED specifically for that link, while extracellular localization remains supported by independent human collagen-II evidence. This conclusion rests on actual body/assay specificity, not title or abstract inference.
Seven exact target rows were read: human HUES1 ECM COL2A1 in the original PMID:23658023(https://pubmed.ncbi.nlm.nih.gov/23658023/) XML Table I and six COL2A1 rows in original PMID:28344315(https://pubmed.ncbi.nlm.nih.gov/28344315/) S2/S3 workbooks. The HUES1 value is 1.2 normalized spectral count; feeder columns are ND. Human bone-marrow specimen categories and aggregated protein identifications remain explicit. Incidental mentions in other protein groups are excluded by exact target matching. Original 28344315 publisher Methods/Results were read in addition to the tables.
The available PMID:28327460(https://pubmed.ncbi.nlm.nih.gov/28327460/) Discussion explicitly identifies collagen II in marrow-MSC matrix, but its original target workbook and complete Methods were not inspected. Exact target tables for 27068509/27559042/28675934 remain unverified. Files downloaded as 28675934 XLSX were HTML and are not treated as data. These source-link limits coexist with independently supported matrix/localization biology. The exact ocular basement-membrane readout remains UNDECIDED. Broad intracellular precursor location is kept non-core and is distinguished from a mature soluble-cytosolic activity.
All 25 original Reactome references were resolved through actual current physical-entity chains to P02458, including multi-collagen complexes and extracellular ligand sets. The original annotations are compartment assertions. Cleavage, receptor signaling and transport activities of other participants are not assigned to collagen just because it is their substrate or ligand. No NEW process is proposed. PAINT source objects remain unchanged; target self-inclusion and short donor lists are not defects. Current same-term donor records corroborate biological decisions but do not reconstruct historical ancestral placements.
The compact source index gives exact source-object decisions, current donor/interaction records, representative Reactome paths with the full expansion pinned, selected-source hashes and access judgments. The owner ledger records 225 actual primary paragraph selections, complete abstracts and exact table rows; bounded independent core selections remain separately attributed. Failed access and provider receipts are retained. Two short cached primary anchors total 5 and 7 words respectively and are not repeated here or in the index. No original machine file, cache, historical record or canonical repository file was edited.
Normal2 validation, ordinary status and rendering passed with seven intentional campaign-policy generic-binding advisories (rows 5–7, 9 and 13–15). The ordinary tool computes DRAFT; that status does not count a campaign completion. All 92 raw GOA rows project to all 92 preserved original assertions, with three products and all 52 original reference pairs intact. The complete audit reproduces 33 literal 42-field positive interaction records, 14 reference-product sequences, 38 current donor joins/187 occurrences, 25 Reactome events/2,040 target paths and seven exact original table rows. The two cached primary anchors total five and seven words.
Since the first complete draft, the YAML changes only its derived status and replaces a short title-like anchor with an equally brief actual abstract clause. Biological decisions, original source fields, products and reference identities are unchanged. The notes correct the aggregate anchor counts and add this validation boundary.
Final read-only source/ownership applicability is pinned to 905a35aeb3efd4a8b4c14248748f25ba8a7e358f with 343 complete opening/closing PR inventories, unchanged-head/base-qualified reuse and actual-current-main merge-base proof for the complete 21,581-path PR 2049. Full queue, project and catalog blobs are authenticated by hash and byte length. All 32 ADD paths are absent at that main, 35 protected paths remain exact, and five shared open-PR source-path occurrences have identical blobs. The complete GO-CAM model and index are separately protected.
Fourteen tooling roots remain unchanged. The scripts root changes only scripts/auto_merge_ready_prs.py, a merge-queue publication-automation script not invoked or imported by the isolated normal workflow; exact old/current tree/blob/diff receipts qualify this unrelated change without relabeling the original snapshot current. The 281 isolated tooling files are authenticated, with only ordinary additive runtime GO-cache entries. ROOT alone checks eventual publisher state, scaffolds history and publishes after independent whole review.
This assessment supersedes the initial row-18 action and the earlier two-anchor count, while preserving the preceding review record. The original IMP/PMID:20603131 object remains present, but its action is now REMOVE specifically because the actual experimental body measures COL1A1/COL1A2 synthesis and secretion in human fibroblasts and mouse Col1a1-mutant cells. Independent human collagen-II incorporation and matrix detection remain positive; they cannot repair the misattributed experimental source. PMID:20603131(https://pubmed.ncbi.nlm.nih.gov/20603131/)
The four binding refinements remain limited to the positive C-terminal-fragment assays and exact tested partners. Official GO definitions classify the protein bound: gelsolin has documented actin-modulating function, FGFR3 receptor-tyrosine-kinase activity, CASP6 caspase-family identity, and GRIN2C an ionotropic glutamate-receptor subunit identity. None requires partner activation, a particular interface, or a native COL2A1 pathway. These claims combine the literal positive pair with established product class; they do not derive an unrelated molecular activity from association. Gelsolin's plasma precursor and CASP6's inactive precursor remain explicitly distinguished from processing states. The seven unrefined positive screen assertions stay non-core under the campaign policy. Comparable assay methods do not make all partner classes identical. PMID:32814053(https://pubmed.ncbi.nlm.nih.gov/32814053/)
The cytoplasm assertion stays non-core. Actual mouse IIA staining leaves some intracellular material unidentified; the paper discusses cleaved N-propeptide uptake as a possibility, while separate ER-lumen annotations concern biosynthetic intermediates. An ER-specific replacement would exceed that source's compartment resolution. PMID:22248926(https://pubmed.ncbi.nlm.nih.gov/22248926/) Homotrimer assembly is positive, but the distinct GO:0070207 process does not settle the original PMID:8660302 homodimerization activity; that source-specific uncertainty remains. Existing parent-term refinements retain distinct source assertions at more informative terms and do not propose new coverage or delete duplicate evidence.
Six brief primary anchors now appear once in YAML, including actual type-I source specificity, human II–IX cross-linking, mouse IIA matrix deposition and the ECM collection method. The exact HUES1 COL2A1 Table I cell (1.2 normalized spectral count) remains separately documented in the original authenticated XML projection; the method quote is not presented as the target-table cell. Total quoted words remain below 25 per source across YAML, notes and index. The original provider failures and provenance appendix remain intact as historical context; no unsupported provider retry or fabricated report was introduced. The biological question now asks how matrix architecture and splice products influence differentiation.