Applied the annotation-reviewer and core-function-synthesizer criteria to all 45 genuinely fetched human FIG4 contexts. Independent annotation-reviewer consultation checked the actual catalytic, rat-donor and complex-assembly evidence. The source objects and order are preserved; the seed supplies no alternative products, and none were invented. No new annotation or publication quotation was added.
The review retains 24 annotations as ACCEPT, 19 as KEEP_AS_NON_CORE, and marks two broad molecular activities TOO_GENERAL. The two core functions are PI(3,5)P2 5-phosphatase activity and protein-serine phosphatase activity on PIKFYVE in human PAS-complex preparations. The evidence document distinguishes the actual assay readouts from proposed mechanisms and records each binding and localization context.
The protein-phosphatase decision rests on actual wild-type/C486S complex comparisons and PIKFYVE phosphosite occupancy, not on a generic Sac-family expectation or a knockout phenotype PMID:33098764. The activity is demonstrated on PIKFYVE serines; broad threonine specificity and isolated-FIG4 kinetic constants remain unestablished. The lipid-phosphatase assay reports phosphate release; positional product assignment is supported by the integrated primary and curated reaction evidence, not by malachite green alone.
The human experimental PI(4,5)P2/PIP3 activities remain non-core because the accessible original abstract supports biochemical substrate breadth without establishing dominant physiological flux PMID:17556371. Rat-donor PI3P/PI4P activity, neurite regulation and recycling-endosome localization are also non-core, with explicit rat/PC12 and abstract-only boundaries PMID:17909536. The absent historical donor rows in current QuickGO do not erase the primary evidence. No new human ER annotation is inferred from the donor's other locations.
Supported generic protein binding remains non-core under the explicit campaign instruction. No kinase-binding subtype is manufactured from the partner name: the PAS studies permit VAC14-mediated association. Exact pair assays were checked where available. The OpenCell PMID:35271311 entry was not individually recovered; its row is retained using independent human PAS-association evidence, while keeping the source-specific verification limit explicit. Similarly, the PMID:16954148 NAS location is corroborated by direct Sac3 evidence without claiming that the PIKfyve-focused source imaged FIG4.
All four R-NUL source reactions explicitly contain human FIG4 within chimeric complexes. Their mouse top-level labels do not justify removing human FIG4 locations. PIKFYVE retains the kinase function and VAC14 the scaffold function [Reactome:R-NUL-1675886; Reactome:R-NUL-1675925; Reactome:R-NUL-1675982; Reactome:R-NUL-1676051].
The disease background is grounded in human CMT4J families and the Fig4 mouse model PMID:17572665, the hypomorphic I41T stability mechanism PMID:21655088, and biallelic FIG4 defects in Yunis-Varon syndrome PMID:23623387. Clinical CMT4J variability PMID:21705420 and familial polymicrogyria PMID:24598713 provide context, not new molecular-function annotations. The reported heterozygous ALS/PLS association PMID:19118816 is not generalized into a uniformly penetrant dominant Mendelian mechanism. Sequence-discovery resources [PMID:9039502; PMID:14574404; PMID:15489334] are background rather than functional assays.
All 31 scoped normal sources were available after acquisition. Seventeen existing main sources remained byte-identical; missing publications and Reactome entries were obtained through the repository's normal fetching workflow. The missing three-row PAINT slice was serialized from verified original IBD data and the complete official family topology, without downloading the bulk leaf-annotation archive. The source guide gives source access and provenance details.
Genuine Falcon research and Perplexity-lite fallback attempts failed with HTTP 403 and 401 responses. No provider report was generated or authored manually. Failed raw provider output remains private and excluded from publication; this review instead records manual synthesis of fetched primary evidence. Cached abstract-only sources were not silently replaced, and no figure-pixel or separate-supplement inspection is claimed.
Targeted schema/reference/ontology validation, strict GOA comparison, PMID checks, read-only status verification, history validation and private HTML rendering passed. The nine annotation warnings concern the explicitly authorized non-core treatment of supported generic binding. The review remains DRAFT under the repository's computed status rules. Conservation checks confirmed all 45 source objects, zero products and zero quotation associations, as well as the qualified source files.
Re-read the primary human VAC14 interaction/architecture evidence and fresh official ontology definitions, synonyms, complete is_a/part_of ancestors and direct paths in response to the review. ROOT reviewed the proposed changes before authoring. The revised actions are 20 ACCEPT, 16 KEEP_AS_NON_CORE, four MODIFY, three MARK_AS_OVER_ANNOTATED and two TOO_GENERAL. The original session record and notes above are retained as history; these are the current dispositions.
The three GO:0006661 rows are over-annotated: the previous reason overread a real broad exact synonym, while the supported human chemistry and separate PI-phosphate branch do not establish synthesis of unphosphorylated PI. Missing ancestry alone is not disjointness. Existing direct dephosphorylation assertions remain accepted; no duplicate replacement was needed. GO:0036092 is valid direct PI3P production through the explicit GO:0043813 part_of edge. It is now non-core and absent from the core-process list to focus that summary on PI(3,5)P2 turnover, without adopting the yeast review's rejection of valid chemical production based on dominant bulk flux.
Four VAC14 contexts now propose scaffold protein binding, supported by each original assay together with independent human binary/architecture evidence [PMID:19037259; PMID:33098764]. Coimmunoprecipitation records remain association assays; binding a scaffold is not performing its scaffold function. Five generic interactions remain non-core under the campaign exception. The serine-phosphatase reasons now point to the ontology's narrow synonym while retaining the demonstrated serine-specific assay boundary. The source guide records official response URLs and hashes.
All 45 incoming source objects/order, 31 normal-source files, two catalytic cores, zero natural alternative products and zero publication quotations are preserved. No NEW annotation was introduced. A new explicit-codex history record documents this follow-up; the previous history is immutable.