The review contains all 61 original annotation contexts and 83 assessed references. The evidence archive preserves the complete original records, source matches, ancestral evidence, canonical sequence, literal interaction pairs and verified excerpt locators. It distinguishes the initial preparation record from the current follow-up.
The single core combines gap-junction channel activity, direct communication and transport, and structural participation in channel assembly. Human Cx32 was distinguished from Xenopus and Neuro2a assay hosts. Selected original material for PMID:10234007(https://pubmed.ncbi.nlm.nih.gov/10234007/), PMID:11891346(https://pubmed.ncbi.nlm.nih.gov/11891346/) and PMID:14627639(https://pubmed.ncbi.nlm.nih.gov/14627639/) supports the stated channel claims; complete original Methods were not recovered for the first paper. Figures and supplements were not independently inspected. Variant gating and reduced abundance remain separate explanations.
The canonical P08034 sequence and direct FASTA agree at 283 residues. No alternative-products declaration is present. PMID:2875078(https://pubmed.ncbi.nlm.nih.gov/2875078/) was read in genuine full original OCR, including Methods, relevant Results, Discussion and captions: human cDNA/translation and antibody recognition are distinct from purified rat gap-junction material. PMID:2460334(https://pubmed.ncbi.nlm.nih.gov/2460334/) original page images show human-sequence peptide antibodies tested on rat junctions. PMID:15383278(https://pubmed.ncbi.nlm.nih.gov/15383278/) is a human Cx32 model fitted to truncated Cx43 density, with explicit loop and boundary limitations.
The authentic HuRI PSI file contains 138 literal target rows. Of these, 132 assay projections map to the 44 current interactions; six extra rows were not assigned by elimination. P08034/ORF4201 has a matching current Ensembl protein cross-reference, but full historical clone-sequence identity was not established. Incoming/source partner suffixes differ for P15692 (-12/-9), Q9BXN2 (-6/-7) and Q9Y666 (-2/-1). All 44 generic-binding assertions are removed for lack of specific functional information. The three product discrepancies remain unresolved and are recorded as SUSPECTED issues requesting CHECK from IntAct, without a corrected accession. Removal preserves the positive assays and does not allege artifacts.
The mouse ortholog sources were traced to actual experimental records. Selected original PMID:20578039(https://pubmed.ncbi.nlm.nih.gov/20578039/) distinguishes Cx32 Western bands from a separate Cx47 immunoprecipitation. Selected original PMID:20089884(https://pubmed.ncbi.nlm.nih.gov/20089884/) shows intracellular and junctional Cx32 in mammary epithelial cells; it does not establish soluble cytosolic residence. Mouse donor extensions were not copied onto human rows.
Current official definitions and typed is_a ancestry support the three nonredundant process facets in one core. The broader cell communication assertion is marked TOO_GENERAL. No NEW annotation is proposed. All three IBA records were matched to actual AN0/PTN008334405 placements and the target AN681/PTN002485527 lineage. Selected human/mouse/rat transmembrane alignment windows support architecture, not every connexin specificity. The original empty IBD taxon field remains literal.
Five official Reactome records and nested entities were read. The GJB1 chain is P08034 residues 1–283, with six copies represented in a connexon. Assembly and biosynthetic compartment roles are distinguished from translation catalysis, generic cargo transport and Cx43-specific mechanisms. The named GO_REF records describe inference methods, not additional experiments.
Reference full_text_unavailable flags describe normal cached records. External original reading is documented separately in each reference assessment. All 79 normal source paths are now present: 71 publications, five Reactome records and three PTHR11984 files. Five inherited paths remain unchanged and 74 were acquired through the native helpers. All 61 annotation contexts and 16 source-export columns were preserved during initial preparation. The cached availability of PMID:14627639 was reconciled to its actual full-text cache; selected original passages were read. Source availability does not imply complete external reading. This follow-up adds no source and changes no cached bytes.
Eleven title-only papers remain UNVERIFIED. The remaining clinical abstracts support bounded genetic context without invented channel assays. Uninspected errata are explicitly recorded, including those linked to PMID:8162049, PMID:10737979 and PMID:15241803. PMID:8733054 has an Arg-to-Trp versus parenthetic R164K inconsistency; no silent variant correction is made. The MGC author correction and other resource metadata do not establish target-specific experiments.
Four short verbatim excerpts appear once each in the review findings: PMID:10234007 (9 words), PMID:14627639 (8), PMID:12460545 (6) and PMID:11891346 (9). The archive contains only hashes and locators for these excerpts. Their authentic PubMed XML-derived abstract locators are preserved alongside all four verified native cache byte offsets. PMID:11891346 requires one whitespace-only line-break adaptation; every quoted word is unchanged. PMID:14627639 uses the exact abstract excerpt repeated in the full-text cache. The archive duplicates no excerpt text, and source-specific access limitations remain explicit.
Initial annotation validation and strict GOA matching passed with no target annotation advisory; the status helper computed COMPLETE. The initial CREATE history, rendering, links and scoped whitespace checks passed. These are historical results for the published parent. The follow-up and genuine EDIT history are now complete. Annotation validation passed without target advisories, strict matching passed for all 61 original contexts, and computed COMPLETE status is unchanged. History, rendering, links and scoped whitespace checks passed. All 79 source files and the ontology cache remain unchanged; these results do not establish publication or integration acceptance. The remaining UNDECIDED action and three suspected product-identity concerns are separate from workflow completion.