ALS2 is the approved human symbol (HGNC:443), accession Q96Q42. The previous symbol ALS2CR6 and UniProt synonym KIAA1563 were checked alongside ALS2 for current-main directories and open review overlap. All were absent at main 1b2db4e9260aedb54abf17fb9fea71c4aee4cc5e; the three canonical seed files match the exact ordinary seed6 recovery/import. The original 48 annotation source objects, qualifiers, WITH/FROM identifiers and three alternative products are preserved. No NEW annotation is proposed.
The substantive core is the long alsin protein's Rab5 exchange activity in endosomal membrane dynamics. ALS2 supplies a mechanistic step, GDP release that permits Rab5 activation, rather than merely being necessary for neuronal survival. Functional self-association is part of this mechanism. Cytosol and early endosomes are the compact core locations; accepted broad or other subcellular locations remain valid without assigning the same biochemical reaction to every observed compartment. No neuronal death, clinical diagnosis or protein-family membership is treated as a new molecular activity.
Official live AmiGO definitions were read. GO:0005096 requires increased GTP hydrolysis, whereas GO:0005085 describes nucleotide exchange. The fully inspected GDP-release source, PMID:15247254, is refined to GEF; the corresponding PMID:12837691 assertion remains unresolved because its complete experiments were not accessible. Likewise GO:0043547 concerns hydrolytic activity, while the human-ALS2 CHO experiment measures Rac-GTP occupancy: the source is refined to positive Rac signaling. GO:0043539 requires kinase binding and activation; downstream PAK1 activity is instead represented by positive regulation of protein kinase activity. These distinctions preserve the positive experiments.
GO:0042803 specifically requires a homodimer. The full self-association paper proposes larger assemblies and competing models, and the later study revises stoichiometry. Generalization to identical protein binding preserves the demonstrated interaction. GO:0032045 describes a functional GEF complex and does not require different subunit species, so the purified/self-associated functional ALS2 unit supports refinement of the generic complex row and the integrated core. No claim of a native ALS2CL heterocomplex is added.
GO:0007032 endosome organization and GO:0016197 endosomal transport have distinct formal scopes; both are supported by the exchange/fusion mechanism. No new process assertion is manufactured from clinical necessity. No matching Q96Q42 activity was found in the local GO-CAM index. No NEW was needed, and no ancestor/descendant duplicate was introduced as a proposed NEW.
PAINT PTN002338289 and PTN002918520 remain ancestral judgments whose exact IBD placement/MSA was not reconstructed. All supplied extant sources are assessed explicitly. Official primary mappings identify mouse Als2 Q920R0/MGI:1921268, rat Als2 P0C5Y8/RGD:1310372, fly Als2 FBgn0037116, and mouse Als2cl MGI:2447532. The exact historical Ensembl peptide version was not recovered. The human target's inclusion is valid experimental grounding, not circularity. There is no argument from donor count or a claimed target-specific catalytic loss. PANTHER PTHR46089 includes Q96Q42, but the machine-generated family description is not used as primary biology.
The normal fresh research attempt used Falcon with perplexity-lite fallback in an isolated temporary output directory; both providers failed before a report was produced (authentication/network failure and DNS, respectively). Existing canonical sources were preserved. Normal GOA-publication caching completed successfully because those identifiers were already cached. Five additional, identified sources were each attempted once through the normal source-contact fetcher and failed DNS: 15033976, 17093100, 17409386, 30224357, 15388334. An earlier wrapper failed before contacting sources because its default uv-cache path was unwritable; that failed wrapper is recorded separately and is not represented as a source retrieval.
There is no provider report or provider-only DOI inventory. All 13 substantive authored PMID references and one Reactome source are included in the census; eight PMIDs and the Reactome record are cached, and the five new records remain required normal-cache gates. Publicly inspected primary papers do not become fabricated publication caches. The original machine alternative-product evidence and untouched UniProt bibliography remain source provenance; uncited bibliography entries are not automatically additional scientific dependencies. External PDF/PMC/DOI links used above map to the same 13 papers.
Initial action census: 30 ACCEPT, 8 KEEP_AS_NON_CORE, 6 MODIFY, 4 UNDECIDED; 48 total, zero PENDING and zero NEW. Four uncertainties preserve source-specific access limits. This is a DRAFT while normal-source gates and any validator advisories remain; it is not a completion or publication claim.
The five required records PMID:15033976, PMID:15388334, PMID:17093100, PMID:17409386 and PMID:30224357 are now available as unchanged normal publication-cache records. Their actual complete abstracts and primary PMID/title/DOI identities were independently checked. All five caches are abstract-only, including the two records with PMC identifiers. The original external full-text reading documented above remains separate from what is present in the cache; this update does not claim to have recovered Methods, figures or supplementary material in those records.
The evidence attachments now cite actual cached results for VPS9-dependent Rab5 exchange, Rac1-dependent ruffle/macropinosome recruitment, Rab5-dependent endosomal fusion, the predominantly tetrameric assembly and GRIP1/surface-GluR2 effects. The ALS2CL record supplies explicitly paralog-level Rab5-binding context. Its human/mouse identities come from PubMed substance indexing, not from species specified in the abstract; HeLa is the host and does not identify every protein insert.
The abstract of PMID:15033976 calls alsin a Rac1 GEF, while PMID:17409386 describes an effector of activated Rac1. The separately inspected original assay details support preserving the positive cellular responses while distinguishing direct Rab5 GDP release from a Rac-GTP capture response. The original results and their qualified disagreement remain intact. Likewise the newly cached tetramer observation in PMID:30224357 strengthens a functional oligomeric core without requiring a universal subunit count. The AMPA-receptor source remains mouse-context evidence; the abstract's summary does not erase the full Results distinction between baseline receptor levels and the AMPA-challenge experiment.
All 48 original annotation objects, their 30 ACCEPT/8 KEEP_AS_NON_CORE/6 MODIFY/4 UNDECIDED decisions, three alternative products and 20 reference identities remain preserved. There is no NEW assertion. All 13 authored primary PMID records and the single Reactome source are now cached, so the earlier five-cache gate is closed. The four source-specific uncertainties still record unavailable original experimental panels or a missing target-specific interaction record, rather than a failed normal fetch. The YAML retains DRAFT with the justified GTPase-activator versus exchange-mechanism advisory; the complete manual audit is ready for independent PR review.
The source-specific exchange decision for PMID:16049005 is now UNDECIDED. Re-reading the full original Methods and Results confirms cellular Rac-GTP capture and a downstream PAK1 assay; direct exchange remains an interpretation. Independent Rab5 studies retain their own accepted GEF annotations. The direct kinase-activator assignment is marked over-annotated, while the observed PAK1 response stays in its existing non-core process row.
All three centrosome annotations are retained as non-core because the supported partial human centrosomal pool has an unresolved contribution to the main endosomal GEF mechanism. This judgment concerns biological context, not the ordinary GOA relationship qualifier. All original source fields, including qualifiers, remain intact. The five other accepted IBA conclusions now use NO_FAILURE_CORE; centrosome IBA and ortholog-transfer conclusions use NO_FAILURE_NON_CORE. Exact historical source/ancestral-node uncertainties remain in source-entity assessments and do not misclassify the accepted target conclusion.
The title and abstract of PMID:12837691 still do not establish the absence of a separate hydrolysis experiment. Its experimental GAP assertion remains UNDECIDED, whereas the fully inspected GDP-release experiment in PMID:15247254 supports refinement. The biological replacement converges on that paper's existing contribution-to-exchange row without asserting monomer sufficiency. Duplicate-target self-binding and Rac-signaling refinements now explicitly describe consolidation. The accurate broad vesicle location is retained with a redundancy note rather than being mislabeled as an unsupported function. The functional ALS2 GEF homooligomer remains supported, with a focused curator question about complex-term scope.
The biological description now omits experimental-access and evidence-ranking commentary. These boundaries remain here: most biochemical data concern the long isoform and engineered fragments; equivalent activities are not established for every splice product, and direct Rac1 exchange is less settled than Rab5 exchange. Source-entity caveats remain self-contained despite some repetition because each annotation's provenance must be interpretable independently.
The follow-up preserves all 48 original source objects, three alternative products, 20 reference identities and the core synthesis. There are no NEW rows or cache changes. Final actions are 26 ACCEPT, 11 KEEP_AS_NON_CORE, 5 MODIFY, 1 MARK_AS_OVER_ANNOTATED and 5 UNDECIDED. The original two published histories are unchanged; this follow-up has a new append-only session record. All 13 PMID caches and the Reactome record remain available. The scientific uncertainties keep the review at DRAFT and are not missing-cache gates.
The second review identified the downstream IC dependency of the Rac-signaling refinement. The original GO:0035022 IC assertion cites GO:0043547 as its premise; replacing that premise by GO:0035022 requires consolidation, rather than retaining an IC assertion whose premise has become its own conclusion. The IC row is now MARK_AS_OVER_ANNOTATED and its hydrolysis-specific premise is SOURCE_BAD. The contextual cellular finding remains represented by the direct IDA refinement: this is an IC-to-IDA evidence upgrade. It does not allege that the historical, unmodified GOA record was self-circular.
The accepted broad InterPro-derived GEF term is classified NO_FAILURE_CORE, while the source-level DH-domain caveat remains UNRESOLVED. The PMID:16049005 GEF row remains UNDECIDED because the inspected original full article's cellular Rac activation does not resolve the scientific disagreement about direct Rac exchange. Access to that article is not the limitation. The PMID:15247254 GEF replacement now explicitly recommends consolidation using the existing contributes_to qualifier; original machine-sourced qualifiers remain unchanged.
Current final census: 48 source annotations; 26 ACCEPT, 10 KEEP_AS_NON_CORE, 5 MODIFY, 2 MARK_AS_OVER_ANNOTATED, 5 UNDECIDED and 0 NEW. Three alternative products, 20 reference identities, all source objects and raw caches, the single Rab5-GEF core and prior histories are unchanged. Two intentional GEF/GAP advisories remain.