EFNB1 source and evidence guide

This guide distinguishes the unchanged local publication caches from separately inspected originals and raw datasets. It supplies locators, not independent experimental evidence. All quoted support fields are omitted; primary reference identifiers and explicit source scopes carry the evidence.

Publication access and exact cache identity

Direct data locators and assay boundaries

• Human urinary-exosome database: EFNB1, NP_004420, one peptide. HTML table index 1, zero-based row 374; the last value 2 is reference 2 (PMID:19056867), not a measurement count.
• Original supplementary dataset for PMID:35922511: 41586_2022_5028_MOESM4_ESM.xlsx, Table 1 row 294, human EFNB1 sequence 1–240; Table 2 cells SO542/U110, PS542/U43 and DU542/U118 are the EPHA2, EPHB2 and EPHA4 reciprocal raw signals. 41586_2022_5028_MOESM6_ESM.xlsx, Figure1_network rows 2/161/162 and Figure2_network rows 112/116/120 preserve the matching records. EPHB2 affinity is marked Published; EPHA4/EPHA2 source values are 25,000/41,000 nM. Excel row numbers are one-based.
• Original U2OS AP-MS network: edge 587044, EFNB1 node 553078 to EPHB2 node 575622, among 39 incident EFNB1 edges. Complete graph input and selected attributes are preserved.
• IntAct target query: literal records retain assay, taxon, host, bait/prey, negative field, features and reference ROGIDs. Curated reference checksum matches are not proofs of full-length assayed insert identity.
• HPA EFNB1 subcellular record: plasma membrane, per-location reliability supported; HPA067188 in HaCaT, PC-3 and U2OS. Images were not independently interpreted.
• PANTHER tree and its saved MSA: human leaf PTN002477887 lies below PTN000885159 and PTN002601663, with an exact 346-residue target match. The official IBD snapshot has six literal inherited source rows; source-specific donor scope is retained.
• Reactome EFNB1 physical entity: P98172 at plasma membrane. All 30 imported event-location assertions are traced to the target in the replay, including complexes and polymer units. Other event participants’ enzyme activities are not assigned to EFNB1.
• The production GO-CAM model already places EFNB1 in ephrin receptor binding at cell surface. It is consistent with the source review; no new process is inferred from other modeled entities.

Products and interpretation

The official displayed sequence is 346 residues. Its three annotated chains are 28–346 (mature), 218–346 (CTF) and 260–346 (ICD), with the short starts inferred. These are processing states, not alternative splice isoforms. The 1–240 tagged extracellular reagent in PMID:35922511, human expression constructs in PMID:16930449 and mouse cortical/tagged-domain work in PMID:19515908 remain source-specific. Positive nuclear localization is not a direct human transcription-factor assay.

All 66 source dictionaries, evidence codes and qualifiers remain intact. The generic-binding policy is the explicit ClinGen curation instruction, rather than a blanket rejection of valid associations. Binding to an established receptor class does not itself assert receptor agonism.

Reproducible calculations

RESULTS.md gives the independently verified partner-class table and measured values. README.md, source-manifest.json and source-projections.json supply exact inputs, URLs, hashes, complete selected fields and the saved-path replay. The dataset-only archive includes no main article text or PDF. Normal caches and external-original availability are reported separately; no existing source flags are rewritten.

Adhesion refinement and interpretation locators

• PMID:8798744(https://pubmed.ncbi.nlm.nih.gov/8798744/), the Abstract in
  publications/PMID_8798744.md, lines 47–56: receptor/ligand cell aggregation,
  inhibition by soluble ligand and the presentation-context comparison. The cache
  SHA-256 remains 8bab9c09663ea065b0e9e9336f7c6b30e5b78a77f2e5fdf55fa26b163bec7133.
  Full Methods remain unavailable. The
  current GO:0098609 record and
  GO:0007155 record identify
  cell-cell adhesion as a direct is_a child of cell adhesion. The refinement
  concerns attachment between the receptor- and ligand-expressing cells, not a
  new process assertion or a universal adhesive outcome.
• Reverse-signaling interpretation uses the existing PMID:16930449 and
  PMID:19515908 locators above, with human-fragment and mouse-context limits
  retained. Positive intracellular localization alone is not a direct
  transcription-factor assay; the integrated extracellular ligand core remains.
• The PMID:35922511 Figure2_network EPHA2 row 120 records 41,000 nM from the
  source's SPR experiment. This is a low-affinity soluble ectodomain interaction,
  not native membrane affinity or demonstrated receptor agonism. EPHA4 row 116
  is also source-measured, while EPHB2 row 112 is labeled as published literature.
• Craniofrontonasal syndrome and X-linked inheritance are recorded in the
  official UniProt entry P98172,
  reproduced in EFNB1-uniprot.txt, CC DISEASE lines 269–276.
  This is an official-record attribution; the cited disease-paper bodies were not
  newly read, and no additional disease mechanism is inferred here.

The existing dataset archive and replay files remain unchanged. They retain the
complete saved inputs used for the reported selections and context denominators;
URLs and hashes identify those snapshots but do not themselves supply their bytes
for offline replay. No article body is added by this clarification.