All 26 current contexts were reviewed individually: 9 ACCEPT, 3 MODIFY, 13 REMOVE and 1 KEEP_AS_NON_CORE. No NEW assertion is proposed. The three refinements specify cell-cell signaling or gap-junction location; the unresolved partner-product correspondence remains explicit as a suspected source concern.
The companion centralizes source-specific observations and access limits. The archive preserves the original contexts and product sequence. There was no prior target namespace or historical review to replace. Clinical associations alone do not manufacture a sensory or developmental process annotation. Generic interaction removal does not negate the observed pair, and partner class alone does not establish a useful GJB3 function.
The canonical product remains the displayed O75712 sequence of 270 residues, preserved in the archive. A canonical-only alternative_products list is omitted because that deprecated slot is intended for multiple products. Experimental reporters are not natural isoforms. Citation metadata for PMID:10757647 includes republication; PMID:9843210 includes an erratum without a PMID. Their contents remain uninspected and no identifier replacement is inferred.
Human Cx31 conducts ions and dyes between transfected cells, supporting electrical and chemical coupling (PMID:16549784(https://pubmed.ncbi.nlm.nih.gov/16549784/)). The original human cloning study identifies keratinocyte expression; its full text remains inaccessible (PMID:9704026(https://pubmed.ncbi.nlm.nih.gov/9704026/)). A mouse donor study reports “Cx31 was localised among type 2 fibrocytes” (PMID:14595769(https://pubmed.ncbi.nlm.nih.gov/14595769/)); this does not establish the same cellular distribution in humans. Human Protein Atlas metadata independently identifies junctional and punctate intracellular pools, with cell-line and antibody details retained in the source excerpts. The puncta are not assigned to a specific organelle.
Research used manual reading of original papers and cached abstracts, the actual PAINT tree/alignment and assertions, official protein and ontology records, original interaction tables, and target-linked Reactome records. The companion distinguishes read main text/captions from inaccessible articles and uninspected images or supplements. No generated provider report or failed provider run is claimed. Disease associations and engineered channel measurements do not alone resolve the tissue mechanism of each pathogenic variant or establish a physiological hemichannel role; no additional process assertion is proposed.