COL10A1

Summary: Type X collagen contributes to the cartilage matrix involved in skeletal development.

Reason: PMID:8554571 reports a segregating human COL10A1 W651R substitution in Schmid metaphyseal chondrodysplasia. That clinical study does not directly measure mutant trimer assembly. The retained developmental role is consistent with the matrix-building protein itself: human NC1 assembly in PMID:11839302 and the positive chick type X matrix/self-assembly observations in PMID:1860888. No new developmental process is inferred from disease association alone. The available clinical and structural evidence supports this broad developmental contribution; it does not specifically establish the narrower conversion of cartilage to bone or the progression of cartilage during endochondral bone morphogenesis. The original TAS term is therefore retained rather than replaced solely from tissue expression.

Summary: The positive COL10A1–BANP pair supports an assay-scoped transcription-factor-binding refinement.

Reason: The exact canonical human BANP product Q8N9N5 and COL10A1 precursor are present in three positive array, prey-pooling and validated Y2H records from PMID:25416956; sequence ROGIDs match the official proteins. BANP is a sequence-specific regulator of RNA polymerase II CGI promoters, independently established for mouse and human in the complete abstract of PMID:34234345. GO:0061629 classifies the bound partner, without asserting that COL10A1 regulates transcription. The screen does not establish a native extracellular/nuclear interaction, mature collagen-trimer binding, a contacted BANP domain or pair-specific orthogonal validation. The refinement combines two inspected evidence types: the cited interaction study establishes the positive pair, while independent functional evidence establishes the bound protein class. It does not infer transcription-factor activity from the interaction itself. Applying the campaign instruction to use an evidence-backed specific term does not require both facts to originate in one paper. GO:0061629 describes the bound class without requiring transcriptional regulation by COL10A1; native compartment separation limits physiological interpretation but does not negate the measured heterologous association.

Proposed replacements: RNA polymerase II-specific DNA-binding transcription factor binding

Summary: Retain the positive canonical UBQLN1 interaction as non-core.

Reason: PMID:25416956 has three exact positive Y2H records for COL10A1 and canonical UBQLN1 Q9UMX0. UBQLN1 is a ubiquitin-binding protein-delivery factor; this pair does not establish binding to the assembled proteasome, ubiquitin itself, or a folding-chaperone activity of UBQLN1. Retain the actual association under the campaign supported-binding policy, without converting a substrate-shuttle identity into an unsupported partner class or native collagen quality-control mechanism. This retention follows the explicit user override in projects/CLINGEN_MENDELIAN.md, Curation instructions; it is not an inference from ActionEnum or a claim that generic binding is a core activity.

Summary: Retain the separate positive UBQLN1 isoform-2 interaction.

Reason: The original PMID:25416956 source names Q9UMX0-2, and its three positive Y2H records match the 561-residue official product. Preserve this separately from the canonical UBQLN1 assertion. Its product identity supports an interaction with a ubiquitin-associated shuttle, not binding to an assembled proteasome or proof that the interaction occurs during collagen secretion. The splice/product boundary remains explicit; no additional specific binding class is established by this assay. This retention follows the explicit user override in projects/CLINGEN_MENDELIAN.md, Curation instructions; it is not an inference from ActionEnum or a claim that generic binding is a core activity.

Summary: Retain the positive interaction with the calcium-sensor protein VSNL1.

Reason: Three positive HuRI records from PMID:32296183 identify the exact 191-residue human VSNL1 product P62760. Its calcium-binding sensor identity is not calmodulin identity and does not justify calmodulin binding or transfer calcium-sensor activity to COL10A1. The Y2H association is retained as non-core, with the tagged COL10A1 precursor and yeast host distinguished from an extracellular collagen complex. This retention follows the explicit user override in projects/CLINGEN_MENDELIAN.md, Curation instructions; it is not an inference from ActionEnum or a claim that generic binding is a core activity.

Summary: The positive MESD pair supports assay-scoped protein-folding chaperone binding.

Reason: PMID:32296183 contains three positive records matching human MESD Q14696 (234 residues) and COL10A1. The official MESD functional record identifies a chaperone that assists folding/transport of LDL-receptor-family proteins, satisfying the partner-class definition of GO:0051087. This is a binding refinement, not evidence that MESD folds COL10A1, that COL10A1 has chaperone activity, or that a mature secretory complex was tested. Precursor sequence and GAL4-fusion features are retained.

Proposed replacements: protein-folding chaperone binding

Summary: The exact NRF1 isoform-4 pair supports RNA polymerase II transcription-factor binding.

Reason: Three positive HuRI records in PMID:32296183 match NRF1 Q16656-4 (522 residues). Its official splice insertion leaves the mapped DNA-binding region intact; the curated NRF1 record establishes a sequence-specific RNA polymerase II transcription factor. This supports GO:0061629 at the tested-product/class level. The insertion overlaps the broader activation region, so no isoform-specific transactivation or native COL10A1 transcriptional mechanism is asserted; the interaction does not localize its contact to that region. The refinement combines two inspected evidence types: the cited interaction study establishes the positive pair, while independent functional evidence establishes the bound protein class. It does not infer transcription-factor activity from the interaction itself. Applying the campaign instruction to use an evidence-backed specific term does not require both facts to originate in one paper. GO:0061629 describes the bound class without requiring transcriptional regulation by COL10A1; native compartment separation limits physiological interpretation but does not negate the measured heterologous association.

Proposed replacements: RNA polymerase II-specific DNA-binding transcription factor binding

Summary: The ZNF410 isoform-3 pair supports a transcription-factor-binding refinement.

Reason: The three positive PMID:32296183 records match Q86VK4-3 (431 residues). The official ZNF410 record describes direct sequence-specific transcriptional activation, and the isoform-3 deletion lies after its five annotated C2H2 DNA-binding fingers. These establish an appropriate RNA polymerase II transcription-factor partner class for GO:0061629. They do not establish the activity of this exact isoform in the Y2H experiment, a particular contacted zinc finger, or a physiological transcriptional role for COL10A1. The refinement combines two inspected evidence types: the cited interaction study establishes the positive pair, while independent functional evidence establishes the bound protein class. It does not infer transcription-factor activity from the interaction itself. Applying the campaign instruction to use an evidence-backed specific term does not require both facts to originate in one paper. GO:0061629 describes the bound class without requiring transcriptional regulation by COL10A1; native compartment separation limits physiological interpretation but does not negate the measured heterologous association.

Proposed replacements: RNA polymerase II-specific DNA-binding transcription factor binding

Summary: Retain the separate BANP isoform-2 interaction with a qualified transcription-factor-binding refinement.

Reason: Six exact positive HuRI records in PMID:32296183, comprising two array/pooling/validated sets, identify Q8N9N5-2 (469 residues). Its verified splice changes retain the BEN DNA-binding domain while altering other regions. The BANP transcription-factor class is supported by the official record and the mouse/human CGI-promoter experiments summarized in PMID:34234345; GO:0061629 therefore describes the bound product class. This does not claim an isoform-2 activity assay, an intact canonical repression interface, a mapped binding domain, or a native COL10A1 nuclear function. The refinement combines two inspected evidence types: the cited interaction study establishes the positive pair, while independent functional evidence establishes the bound protein class. It does not infer transcription-factor activity from the interaction itself. Applying the campaign instruction to use an evidence-backed specific term does not require both facts to originate in one paper. GO:0061629 describes the bound class without requiring transcriptional regulation by COL10A1; native compartment separation limits physiological interpretation but does not negate the measured heterologous association.

Proposed replacements: RNA polymerase II-specific DNA-binding transcription factor binding

Summary: Retain the SGTB interaction without converting a cochaperone label into chaperone binding.

Reason: Three positive PMID:32296183 records identify the exact human SGTB product Q96EQ0. Its official record describes association with HSC/HSP70 proteins and regulation of their ATPase activity, with relevant functional statements inferred by similarity. SGTB is not itself an Hsp70 protein, and this pair does not establish that it supplies the folding/transport activity required by the chaperone partner-class definition. Keep the positive association non-core without assigning COL10A1 cochaperone activity. This retention follows the explicit user override in projects/CLINGEN_MENDELIAN.md, Curation instructions; it is not an inference from ActionEnum or a claim that generic binding is a core activity.

Summary: Retain the positive CIDEB interaction as non-core.

Reason: The exact 219-residue CIDEB Q9UHD4 product has three positive HuRI records in PMID:32296183. Its official lipid-droplet fusion/lipid-transfer role does not establish chemical catalysis or make it an enzyme merely because material is transferred. No supported enzyme-binding refinement or COL10A1 lipid-droplet role follows from these records. Preserve the positive Y2H pair and its source scope. This retention follows the explicit user override in projects/CLINGEN_MENDELIAN.md, Curation instructions; it is not an inference from ActionEnum or a claim that generic binding is a core activity.

Summary: Retain the positive UBQLN2 interaction as non-core.

Reason: Three positive PMID:32296183 records match canonical human UBQLN2 Q9UHD9. Its ubiquitin-binding delivery and condensate roles do not mean that the tested isolated partner is the assembled proteasome or an autonomous folding chaperone. The exact association is preserved without importing proteostasis or condensate-scaffold functions into COL10A1. This retention follows the explicit user override in projects/CLINGEN_MENDELIAN.md, Curation instructions; it is not an inference from ActionEnum or a claim that generic binding is a core activity.

Summary: Retain the independent HuRI canonical UBQLN1 assertion.

Reason: Three positive records in PMID:32296183 match canonical Q9UMX0. This is distinct from the PMID:25416956 canonical and isoform-2 source assertions and is not collapsed into either. The source supports the Y2H association but does not establish proteasome-complex binding, contact with ubiquitin, or a native collagen-delivery function. This retention follows the explicit user override in projects/CLINGEN_MENDELIAN.md, Curation instructions; it is not an inference from ActionEnum or a claim that generic binding is a core activity.

Summary: Reactome places the COL10A1-derived substrate outside the cell during collagen degradation.

Reason: The R-HSA-1454757 input hierarchy resolves through cleaved type X collagen to UniProt Q03692 in the extracellular region. This supports the compartment of the collagen substrate, not a protease activity of COL10A1. Human ECM-enriched proteomics independently supports the extracellular assignment.

Summary: Secretion places assembled type X collagen in the extracellular region.

Reason: The R-HSA-2089971 physical-entity hierarchy contains type X collagen chains/trimer resolved to Q03692 on the secretory and extracellular sides. This is a curated compartmental model, corroborated by human ECM proteomics and type X extracellular matrix assembly; it is not an experiment demonstrating a trafficking enzyme activity of COL10A1.

Summary: Type X collagen is an extracellular network-forming matrix component.

Reason: R-HSA-2213207 resolves the type X collagen entities to Q03692 in the extracellular region. The complete original chick abstract in PMID:1860888 reports matrix-associated material and purified collagen self-assembly; the in-vitro hexagonal lattice is not asserted as the exclusive native human architecture. The current GO collagen-type-X term explicitly distinguishes in-vitro lattices from in-vivo fibril association.

Summary: Association with collagen type II occurs in the extracellular matrix model.

Reason: The R-HSA-2213208 hierarchy directly identifies type X collagen chains as Q03692 in the extracellular region. Retain the curated location of that collagen association; do not treat the model as a new direct human binding experiment or transfer the complete type II fibrillar mechanism to type X collagen.

Summary: The modeled type X collagen ligand is extracellular.

Reason: R-HSA-2465890 resolves its collagen-ligand set through type X collagen to Q03692 outside the cell. This supports ligand localization, not DDR2 kinase activity in COL10A1 or proof of an isolated human COL10A1–DDR2 pair in the current source set.

Summary: Refine the broad collagen-trimer annotation to the type X homotrimer.

Reason: PMID:8554571 supplies human COL10A1 disease and domain context, rather than a direct trimerization assay. Independent recombinant human NC1 structure in PMID:11839302 establishes a trimer, and the official type-X definition identifies alpha-1(X) homotrimers. GO:0005599 is therefore a justified specificity refinement while retaining the original TAS source object and its narrower experimental scope.

Proposed replacements: collagen type X trimer

Summary: Retain inherited type X collagen homotrimer membership.

Reason: The source includes COL10A1 itself, which is legitimate target experimental grounding for a PAINT inference, not circularity. Current matching records include the direct IPI assertion from PMID:11839302, and the recovered human NC1 structure supports trimerization. The historical PTN008601431 tree/MSA and ancestral-node placement were not recovered; current source mapping does not replace that phylogenetic judgment or provide evidence against it.

Summary: The recombinant human NC1 domain forms the type X collagen trimeric assembly.

Reason: PMID:11839302 directly determines a 2-angstrom structure of human C-terminal NC1 produced with an N-terminal His6 tag in E. coli TG-1. Residues 549–680 are resolved in the trimer; the N-terminal 28 residues of the 161-residue construct and the tag are disordered. This is direct domain-level assembly evidence consistent with the native type X homotrimer, without claiming a full-length human collagen network was reconstituted.

Summary: Retain ER-lumen localization during collagen proline hydroxylation.

Reason: R-HSA-1650808 resolves a collagen substrate entity to COL10A1 Q03692 in the ER lumen. The prolyl hydroxylase catalyzes the modification; COL10A1 is the luminal secretory substrate. This biosynthetic compartment is retained but does not define its mature extracellular structural core.

Summary: Retain ER-lumen localization during P4HB-associated collagen maturation.

Reason: R-HSA-2002460 resolves the collagen chain substrate through the current physical-entity hierarchy to Q03692. Its luminal association with P4HB is a biosynthetic context, not an oxidoreductase or chaperone activity assigned to COL10A1. The mature structural role occurs outside the cell.

Summary: Retain ER-lumen localization while type X collagen chains assemble.

Reason: R-HSA-2022073 includes COL10A1 Q03692 in the procollagen assembly hierarchy. This supports residence of the assembling secretory protein in the ER lumen; it does not mean that the mature matrix activity is performed there or that every generic procollagen processing detail is specific to type X.

Summary: Retain the ER-lumen side of the collagen secretion model.

Reason: R-HSA-2089971 explicitly represents COL10A1-containing collagen entities before secretion as well as extracellular products. The ER location is appropriate for biosynthesis and export, whereas the integrated functional core is extracellular matrix assembly. This separate source assertion is preserved alongside the extracellular-region row.

Summary: Retain luminal type X procollagen association as a biosynthetic location.

Reason: R-HSA-8944262 resolves the type X procollagen entity to Q03692 and places its association in the ER lumen. This curated assembly compartment is consistent with a secreted collagen, but it is not a second autonomous ER molecular activity or an enzymatic processing role of COL10A1.

Summary: Retain the collagen structural classification, separating proteomic detection from mechanical measurement.

Reason: The original human bone-marrow study PMID:28344315 identifies COL10A1/Q03692 in its authentic Supplementary Table S3 after insoluble ECM enrichment and LC-MS/MS. The study does not measure isolated collagen-X tensile mechanics. The current GO:0030020 comment explicitly permits extracellular matrix collagen proteins to receive this term; type X has independently supported matrix assembly, so retain the curated classification with type-X-specific scope. It does not assert a canonical fibrillar-collagen mechanism.

Summary: Retain the inherited extracellular collagen structural role.

Reason: Current source tracing identifies human COL4A1 and C. elegans LET-2 with matching experimental or curated functional grounding. Donor number is not the basis of the PAINT inference, and PTN001166351 historical node placement remains uninspected. The official GO:0030020 collagen convention and COL10A1-specific matrix/assembly evidence support retention without equating type X with every donor collagen architecture.

Summary: Retain the matrisome-derived structural classification with explicit assay scope.

Reason: PMID:25037231 Additional file 2 identifies COL10A1 as a core-matrisome collagen in human ECM-enriched tissues, with 41 aggregate unique peptides in the complete data set. RCA here is classification plus abundance/detection evidence, not a tensile test. The current GO:0030020 comment permits extracellular collagen annotation, while independent type X domain and chick assembly evidence supports a structural contribution. Preserve that classification without claiming measured full-length human tensile performance.

Summary: COL10A1 contributes matrix structure during extracellular matrix organization.

Reason: Human NC1 assembles into trimers, and the complete original chick abstract in PMID:1860888 describes type X matrix-associated material and purified self-assembly. COL10A1 therefore supplies organizing structure rather than merely being a substrate of another matrix enzyme. Current source identities and same-term collagen records are traced separately; neither numerous donors nor a presently missing same-term row reconstructs PTN002770636. No target-specific loss of the inherited organizing role is established.

Summary: Retain the extracellular matrix location with independent target evidence.

Reason: The selected PMID:17876790 cache is a collagen-family review abstract; its exact COL10A1 passage was not accessed, and the original EXP source object is preserved without pretending it was verified from that abstract. Human COL10A1 target rows in the original ECM-enriched proteomic supplements (PMIDs:25037231 and 28344315), together with type X assembly evidence, independently establish the location. This supports retention and curator deference, not a wrong-paper claim.

Summary: Human COL10A1 is detected in an ECM-enriched tissue preparation.

Reason: The authentic PMID:25037231 Additional file 2 identifies COL10A1/Q03692 in the collagen division, including normal-colon samples. The Methods remove cellular fractions before digestion and LC-MS/MS of insoluble ECM-enriched material. This supports matrix localization in mixed human tissue; it does not resolve a producing cell type or measure a mechanical function of the isolated protein.

Summary: Human COL10A1 is identified in insoluble bone-marrow ECM-enriched material.

Reason: PMID:28344315 Supplementary Table S3 contains the exact human Q03692/COL10A1 target, with six distinct peptides across the complete data set and detections in healthy-donor and one newly diagnosed myeloma preparation. Table S2 uses a separate cohort-level signature classification; it should not be read as proving no myeloma detection. Aspirate extraction and LC-MS/MS support the ECM location but not a cell-specific source, purified collagen mechanics or universal disease abundance.

Summary: Retain the inherited extracellular matrix site of collagen-X structural work.

Reason: Direct human ECM-enriched proteomics and type X assembly evidence support the target location. Current mappings identify the listed collagen donors and trace their actual same-term records, including legitimate target self-grounding. Some current source mappings lack an exact term or contain only inferred annotations; those are recorded as present-day limits, not evidence that PTN008390299 was placed incorrectly. The historical PAINT tree/MSA was not inspected, and no target-specific loss of matrix localization is established.

Summary: The isolated recombinant human NC1 trimer binds calcium.

Reason: PMID:11839302 provides a 2-angstrom structure with a buried calcium cluster and independent inductively coupled plasma analysis after Chelex-treated-buffer dialysis, with an EDTA pre-dialysis comparison. The protein is the human C-terminal 161-residue NC1 construct carrying an N-terminal His6 tag, produced in E. coli; the result is not a full-length/native-matrix calcium stoichiometry or mineral-nucleation assay. GO:0005509 has no ancestor/descendant overlap with any of the 34 seeded terms; its two narrower synaptic-buffering descendants do not fit these assays. This is a secondary binding capability, not another structural core or a new process. UniProt Q03692 records calcium-coordinating sites at residues 626, 627, 633 and 634 and cites this same study and PDB 1GR3; these features corroborate the structural assignment but are not a second independent calcium-binding experiment.