FMR1 source guide

This guide accompanies the review, journal, and structured source evidence. It records what was actually available and what the experiments establish. Primary-source reading was divided among Codex reviewers, followed by cross-review of all four annotation groups; this does not mean every reviewer read every original paper.

Context and source preservation

The genuine fetch produced 253 original QuickGO records and 250 stored annotations. Three projection pairs (raw 33/34,174/175,235/236; zero-based) share exported seed keys. Both originals remain in the evidence file. All original terms, evidence codes, references, qualifiers and partner fields are retained. Eight original extension groups contain 12 conjunctive atoms and are represented through the schema's extensions slot, preserving relation/target/order. The two RNAcentral targets are synthetic structural RNAs, not endogenous human transcripts. The neuroblast extension retains CL:0000031, whose preferred label is neuroblast (sensu Vertebrata).

The source set contains 85 immutable normal files:82 publications and three PTHR10603 files. Eighteen existing main files were hydrated before genuine fetch and remain exact. Additional source qualification and genuine acquisition expanded the initial 69-file set to85. A successful full-text download is not a claim that its complete content was read. The six later donor publications are retained for checkable provenance; their newly acquired bodies were not silently promoted to inspected experiments. The initial source-access map remains a dated 69-source observation, superseded in scope by this 85-source inventory.

Manual source research was used after known provider authentication/quota failures in this campaign. No generated-provider report is fabricated. Supporting entries contain reference IDs without new publication quotations, following the standing ClinGen project instruction. That instruction also governs generic binding: supported generic associations remain non-core, supported specific classes are refined, and unresolved source contexts remain undecided. This is an explicit user instruction, not a change to the default validators.

Biological decisions and assay boundaries

The 250 stored decisions are 81 ACCEPT, 139 KEEP_AS_NON_CORE, 8 MODIFY, 6 TOO_GENERAL and 16 UNDECIDED. Three distinct core activities cover selected-mRNA recognition, translation regulation and molecular-condensate scaffolding. Developmental, viral, nuclear/chromatin and channel effects are retained with source-specific scope rather than treated as universal extensions of those cores.

The positive AMPA-receptor-internalization annotation from the original 2015 source is refined to negative regulation of internalization. The measured rescue reduces the elevated internalized/total-receptor signal; this is not an isolated rate-constant measurement. The 2014 chromatin paper supplies separate corroboration and is not substituted for the original 2015 citation. Positive filopodium assembly is generalized to regulation of assembly because the donor's combined filopodia/spine protrusion assay does not isolate a positive assembly rate. The presynaptic-translation IBA remains undecided: its phylogenetic placement is genuine, but the inspected channel/exocytosis study does not measure local translation. This is an access/mechanistic limit, not a donor-count argument or an asserted ancestral loss.

FMRP can repress or enhance translation depending on the RNA and regulatory system. RNA structure, translation, stability and localization were measured separately and are not interchangeable. The short-RNA annealing experiments do not establish universal RISC necessity; the FMRP-deficient-cell study provides relevant counterevidence to that universal claim. Chromatin recognition and damage-response recruitment are not catalytic DNA repair. A source's mRNA localization or protein association is not automatically evidence for direct protein contact.

Native human assays, tagged human protein in rodent hosts, rodent proteins in human hosts, and cell-free fragments are distinguished. The human FMRP 445–632/CAPRIN1 607–709 condensate experiments do not assay every full-length natural product. CNOT7 performs the deadenylation in that system. The 2024 FXR1 study includes real FMRP network experiments, but its actomyosin-signaling conclusion is not transferred to FMRP. Its FMR1 knockdown did not reproduce the tested A549 stress-fiber phenotype.

Natural products

All 11 seeded products are preserved. Alternative-sequence edits from the official UniProt JSON reconstruct every official product FASTA exactly. The displayed 632-residue Q06787-1 is named 6. Historical paper isoforms, names such as ISO6/12, and UniProt accession suffixes are different naming systems; matching numbers is not a sequence mapping. The record says at least 12 products while enumerating 11; no unenumerated twelfth product was invented. Artificial fragments are not additional natural products.

Phylogenetic evidence

The actual PTHR10603 tree and MSA each contain 149 entries. The human target PTN002471935 reconstructs the 632-residue displayed sequence exactly. All 11 incoming IBA origin/term joins occur on its actual path. The 11-record normal PAINT slice is derived from the original IBD GAF, whose URL, retrieval time and hash are preserved in the evidence file, alongside the official tree/MSA receipts. PTN001318100 and PTN000063980 are ancestral nodes; they are not individual pairwise donors.

Tree/MSA and IBD are separately pinned release snapshots. In particular, nucleus raw 156 has different donor/date fields between original QuickGO and original IBD; neither is rewritten. Target self-membership and a short donor list are not treated as circularity or weak evidence. Node placement establishes the propagation path; it does not replace biological interpretation of the asserted function.

Publication access and assessment

“VERIFIED” means the identifier and intended supporting claim were checked to the stated scope. It does not imply a full-paper or supplement audit. “UNVERIFIED” identifies an unresolved claim-specific assay, target record or donor-paper read; it does not imply a wrong paper or a false biological interaction. Official donor records establish curated provenance, not unseen experimental detail. The cache metadata flag is preserved independently in the evidence file.

Reproducible evidence

The accompanying evidence YAML contains all 253 original source objects, the exact raw-to-stored projection, original extension groups plus typed serialization, all 11 sequence reconstructions, all 11 IBD joins, and all 85 normal-source hashes. Official entry points are UniProt Q06787, QuickGO, and the recorded PANTHER API/GAF URLs. This public evidence is self-contained; private audit packets are not required to recover original source contexts.

The original 18539120 author-lab PDF was read through cached web extraction, while a direct subsequent request returned 404 HTML. That failed HTML response was not treated as a PDF or supporting evidence. Selected Methods and Results supplied the transport/motor scope; no supplement, movie or figure-pixel inspection is claimed for that paper. Other paper-specific figure/SI access is stated in the table and preserved in the source-review notes.

Content-review follow-up

Translation regulation is now an explicit molecular-function core (GO:0045182), covering target-dependent repression and enhancement. The separate mRNA-binding core retains recognition, stability and transport; the condensate-scaffold core is unchanged. The selected original ASCL1 and Sod1 experiments distinguish human-cell and mouse/STEK evidence, RNA abundance from translation, and tested constructs from untested natural products. The SoSLIP element's IRES-like property is FMRP-independent and is not assigned as an additional FMRP activity.

The two PMID:31439799 UNDECIDED decisions assess complete source assertions. One combines established mRNA binding with a direct positive regulation relation to signaling-adaptor activity. The other combines signaling-adaptor activity, CAPRIN1 input, negative translation regulation and a direct positive regulation relation to condensate scaffolding. All five original extension atoms in those two contexts remain intact. Established binding and condensate results are accepted elsewhere; the unresolved causal conjunction is neither a claim that those activities are absent nor a reason to discard the original relations. The paper's discussion presents cellular-signaling integration as a prediction. CNOT7 performs the measured poly(A) cleavage.

The PMID:18664458 cytoplasm assertion is retained as non-core. Its positive observation—endogenous FMRP recruitment into stress granules after TDRD3 overexpression—remains explicit, without replacing the broad source context with an exclusive or constitutive stress-granule location. Independent stress-granule contexts remain in the review.

Five reference entries now carry concise, quote-free findings for the central translation and condensate studies (PMIDs 11157796, 19097999, 19166269, 30765518 and 31439799). Their access flags and statements distinguish authentic abstract reading from the separately inspected original primary text. No cached publication, product record or original annotation field changed.