The review retains all 53 imported source contexts, including the four historical assertions absent from the current annotation feed. The current GOA contains 49 contexts; the exact earlier 57-row file is preserved in the dated source snapshot. Changes in feed coverage are provenance changes, not automatic biological rejections or retirement decisions. The distinct PMID:29599245 source is not replaced by PMID:29599213.
The integrated core describes delta's structural and regulatory contribution to eIF2B exchange and cytoplasmic translation initiation in cytosol. Positive human assembly/interface and patient-extract evidence is distinguished from hamster perturbations, the S. pombe structural preparation and whole-complex assays. Imported enables relations remain unchanged while three source-specific queries ask whether contributes_to better represents the evidence; these are conditional checks, not confirmed errors or an isolated-subunit activity requirement.
The native HeLa preparation in PMID:11323413 resolves cytoplasm. Target-resolved Reactome component 72521 independently supports the finer cytosol location. The T-cell stimulation results in PMID:8626696 show downstream whole-complex activation but leave delta's work in the T-cell receptor signaling process unresolved. PMID:15060152 reports negative isolated beta associations alongside positive multicomponent assembly; factor-binding refinements also cite independent human interface evidence.
The positive EIF2B4 clinical evidence in PMID:14566705 is kept separate from the negative isolated premature-ovarian-failure screen in PMID:15507143. Rat heat, glucose and peptide-hormone donor annotations retain their organismal and whole-complex limits. Unresolved clinical/developmental assertions are not called wrong-gene annotations merely because complete supporting evidence could not be recovered.
Three natural products of 523, 543 and 522 residues are preserved. The overexpressed 543-aa Flag construct called V1 in PMID:37014850 is not assigned to a UniProt product by length alone. Its free-delta sequestration and altered gamma/epsilon subcomplex models remain alternatives, not established autonomous delta exchange mechanisms. The product-comparison questions describe a testable uncertainty rather than a new process annotation.
Supported generic binding is retained as non-core when no evidence-backed narrower class is available, following the ClinGen project policy. Verified partner classes support refinements without requiring purified binary interfaces or transferring the partner's activity to delta. Reference-sequence checksums and assayed regions remain separate.
Three inherited publication supporting_text fields were removed after exact source checks: the noncontiguous PMID:8626696 excerpt and two copies of an unsupported PMID:29599213 paraphrase. Their reference IDs remain. All other inherited quotation occurrences are preserved. The retained generated-research excerpts are not upgraded to independently verified primary evidence.
The source guide gives the actual access scope, original locators and cache hashes. HTTP failures and no-result responses are retained in provenance; selected indexed originals are not described as complete papers. Normal publication flags remain exact. The reproducible analysis derives literal database joins from archived inputs without republishing article bodies.
This dated decision supersedes the earlier provisional statement that all four historical absences were retained without retirement. The complete fresh QuickGO query returned 49 annotations on one page and matches all 49 contexts in the normal-current GOA, including evidence, references, qualifiers and source identifiers. Historical indices 2, 6 and 7 (GO_REF:0000043) and index 13 (PMID:29599245) are now explicitly marked retired. All 53 review objects remain: 49 active and four retired. Retirement records disappearance from the current source snapshot, not a biological REMOVE judgment; the original source identities, scientific reviews and dated old GOA snapshot are preserved. PMID:29599245 remains distinct from PMID:29599213.
The clinical PMID:12499492 heat-response row is retained as non-core using independent corroboration from the rat Q63186 IDA evidence in PMID:9341116. That study's authentic abstract describes heat-dependent regulation of whole eIF2B in rat H35 cells. It does not identify delta as a temperature sensor or supply a direct human delta perturbation. The clinical paper remains abstract-only, its exact delta-specific evidence is unresolved, and its inherited supporting text is unchanged. Corroborating the biological annotation does not claim recovery of the clinical paper's complete experiments.
The revised disposition counts are 21 ACCEPT, 18 MODIFY, 12 KEEP_AS_NON_CORE and two UNDECIDED across the 53 retained objects. All 83 supporting_text occurrences, the integrated core and three product objects are unchanged.
This section supersedes the earlier dispositions only for source rows 9, 12, 36, 46, 47, 48 and 51. Row 12 retains the positive PR8 M2/EIF2B4 AP-MS and co-IP association as KEEP_AS_NON_CORE. Identifying M2 as a channel does not exclude an intervening complex component, so the transporter-binding replacement is withdrawn. Rows 46/47/48/51 also become KEEP_AS_NON_CORE, preserving the literal validated binary two-hybrid interactions. The independent FOS/KLKB1/LPL/GATM classes remain contextual evidence; exact experimental inserts, processing states and natural-product equivalence are unresolved, and no independently established EIF2B4 role with these partners was recovered. The specific replacements are withdrawn without declaring false hits or using different native compartments as disproof. The established intracomplex initiation-factor-binding refinements remain unchanged.
Rows 9 and 36 become MARK_AS_OVER_ANNOTATED. PMID:14566705 provides positive EIF2B4 multisystem disease and, for the A391D patient, ovarian dysgenesis. PMID:15507143 is a negative known-variant screen of 93 isolated-POF patients and discusses earlier ovarian disease and a proposed apoptotic explanation. The narrower process assertions are not established by these clinical observations alone. No wrong-gene, confirmed wrong-reference or absence-of-function conclusion is made.
Heat-response rows 29 and 30 remain non-core, with reference-only support from the actual rat Q63186/PMID:9341116 donor evidence. PMID:8567668 and PMID:8430778 are likewise attached directly to the glucose and peptide-hormone rows 33 and 34. The authentic abstracts and exact donor joins support the stated whole-factor contexts; complete operative donor texts and the clinical paper's precise delta heat experiment remain unavailable. Generic inherited excerpts are not treated as donor-specific evidence.
After individual evidence review, 23 duplicate title excerpts and three mismatched background excerpts are removed. Three titles that directly support their narrow findings remain, as do the other 54 inherited excerpts. The three mismatched row-local support entries are also removed while their top-level references and other uses remain. The resulting 57 excerpts are unchanged; no publication quotation is added, replaced or relocated. Reference-only anchors and the source guide locate the evidence. The supported-generic-binding policy is recorded at CLINGEN_MENDELIAN.md, line 2976. Broader translation regulation and positive initiation describe one integrated core; the contributes_to relations in rows 37 and 41 remain explicit and unchanged.
The resulting totals are 21 ACCEPT, 13 MODIFY, 15 KEEP_AS_NON_CORE, 2 UNDECIDED and 2 MARK_AS_OVER_ANNOTATED across the same 53 source objects, including the four already retired historical assertions. The three products and single core are unchanged. The database replay still reports literal source identities and joins, not these scientific dispositions. The exact approved reference-only renderer correction from PR #5285 is included, so absent excerpts display as evidence references without empty blockquotes. The archive redesign and new-quotation suggestions remain outside this follow-up.
This entry supersedes the earlier non-core disposition only for row 8, the ARBA00026561/GO:0007417 CNS-development inference. The saved rule really matches the delta FunFam; there is no wrong-family claim. The positive EIF2B4 patient findings remain recorded, but the family assignment and clinical consequences do not identify a CNS-development step beyond delta's contribution to translation. Row 8 becomes MARK_AS_OVER_ANNOTATED, as does the already assessed organ-development inference in row 9. This is an assessment of the two actual rules and source scopes, not a transfer of another subunit's review. It does not change the separate experimental myelination row or assert absence of all developmental involvement. Both electronic rows now include a structured term-scoping diagnosis with their exact rule identifiers.
The accepted GO:0031369 ISS row 27 now cites PMID:9446619 directly in supported_by. Its rat-liver far-Western evidence was already assessed and remains distinct from human complex corroboration; no native isolated human pair or new assay is claimed. PMID:14566705 is likewise attached directly to row 8. The existing reference assessments and original-access limits remain unchanged.
The request for additional publication quotations is not adopted under the standing zero-new-quotation instruction. All 57 inherited excerpts remain exact, with reference-only anchors and the source guide providing traceability. The resulting counts are 21 ACCEPT, 13 MODIFY, 14 KEEP_AS_NON_CORE, 2 UNDECIDED and 3 MARK_AS_OVER_ANNOTATED across 53 preserved objects (49 active and four retired). Three natural products, the single core, all source identifiers and qualifiers, and all other review decisions remain unchanged.