DIAPH1 is HGNC:2876 and UniProt O60610. The normal isolated fetch produced 60 annotation rows from 62 raw QuickGO results: two pairs differ only in source bookkeeping fields and map to the same seeded annotation identity. All 60 imported dictionaries, including evidence, reference, qualifier and supporting entities, are preserved. The three official protein products (1,272, 1,248 and 1,263 residues) are retained. Historical construct lengths and gene-level proteomics do not establish equivalent assays of every product.
The draft contains 14 ACCEPT, 30 KEEP_AS_NON_CORE and 16 MODIFY decisions, one core function, 38 references and no new annotations. The evidence guide records source access, actual assays, partners and propagation paths. Broad organizational terms are refined to actin filament polymerization; broad cytoplasmic, cytoskeletal and spindle locations receive supported specific replacements. Existing contextual locations and processes remain visible.
All eight generic protein-binding rows have a supported partner class: small GTPase binding for RHOA/RHOC/RHOD, protein phosphatase binding for POPX2/PPM1F, signaling receptor binding for RAGE, and protease binding for the engineered HIV retropepsin association. The latter retains the D25-mutant bait limitation and does not establish wild-type physiological binding. The project’s explicit curation instruction supports evidence-based retention and refinement; no interaction is removed merely because its source term was generic.
The independent core consultation was assessed against the original assay scopes. One integrated Rho-regulated actin-assembly unit is retained. The platelet-cytosol assay is not a completely purified actin reconstruction; its human fragments and the human R1213* cellular experiments provide complementary evidence. RhoD binding is explicitly weak in the original experiment, whereas RhoA/B/C restore assembly activity. DAAM1-specific thrombin co-immunoprecipitation is not assigned to DIAPH1. The microtubule and calcium-response contexts do not create separate catalytic cores.
The RNA-capture table explicitly contains DIAPH1, and the RAGE and POPX2 originals were read. PKD2 work separates human endogenous evidence from the mouse screening library and porcine imaging. Mouse brush-border proteomics was recovered as genuine full cached text; this supports the inherited location without establishing the proposed tip-complex mechanism. Nuclear, neuronal and synaptic donor evidence is retained with access and product limitations described in the guide.
The exact PAINT ancestral node and target descent were inspected. One historical zebrafish seed identifier remains unresolved in the current family tree; the named mouse and worm descendants and independent human evidence supply positive grounding. The unresolved identifier is not replaced from memory and is not treated as a count-based objection to the ancestral assertion. Reactome locations were traced through actual component/member relationships to O60610, including secretory and ficolin-rich granule membranes and their plasma-membrane output entities.
The configured Falcon research attempt returned an authentication error; the configured Perplexity fallback returned insufficient quota. No provider report was invented. Existing machine caches remain byte-identical. Separately recovered originals do not change their abstract-only cache flags. One initially reused purported BioPlex HTML was a challenge page; it was excluded and replaced only in the additive evidence collection by a separately verified genuine PMC original. This did not alter any publication cache.
Two unsealed draft wording corrections followed independent review: the top-level summary now describes alternative-splicing biology without curation commentary, and the RHOC BioPlex rationale now states RHOC-specific corroboration. A source URL in the guide was corrected against the saved retrieval receipt. The original draft hash and exact changes are recorded with the review evidence.
This entry supersedes the earlier protease-binding refinement and its statement that all eight generic binding rows had justified specific replacements. The exact PMID:22190034 records and original AP-MS workflow support DIAPH1 association with an engineered D25-mutant HIV retropepsin bait. They do not establish direct protease recognition or a native protease-regulatory mechanism. The literal GO:0005515 source row is therefore retained as KEEP_AS_NON_CORE under the explicit project instruction. Its source dictionary is unchanged. The other seven binding refinements retain their independent recognition evidence.
The three screening references now have MEDIUM relevance and distinct source-specific notes. The HeLa RNA-capture paper contains a verified DIAPH1 Table S1 hit, with original Methods assigning unique peptides to gene models; it is supporting target evidence, not a passing mention. Its scope remains RNA contact without a particular substrate, binding domain or product-specific regulatory mechanism. HIV AP-MS and BioPlex provide association evidence. PMID:20937854 remains HIGH because its original directly perturbs mDia1 in human cells and measures GSK3/microtubule responses; its precise experimental and access limits are now stated in the reference review.
Final counts are 14 ACCEPT, 31 KEEP_AS_NON_CORE and 15 MODIFY, with the same 60 source objects, 38 references, three products and one core. No NEW annotation or quotation is added. The retained generic-binding advisory makes the normal computed status DRAFT; this records review status honestly while preserving the supported association. The optional core-specificity and expanded-findings suggestions do not justify speculative changes to the accepted core or protected caches in this bounded follow-up.