log/R25.md — Round 25 entries

Sub-file of log — see parent for index.

[2026-05-07] verify | interventions/pharmacological/telomerase-activators.md

Pages verified: 1

interventions/pharmacological/telomerase-activators.md — corrections: 8 (see below)

PDFs read:

Townsley 2016 (10.1056/NEJMoa1515319) — downloaded this pass (bronze OA via camoufox); verified end-to-end (10 pp.)
Khincha 2018 (10.1182/bloodadvances.2018016964) — downloaded this pass (gold OA); verified end-to-end (7 pp.)
Bodnar 1998 (10.1126/science.279.5349.349) — local; verified end-to-end (4 pp.); consistent with tert and replicative-senescence verified pages
Bernardes de Jesus 2011 (10.1111/j.1474-9726.2011.00700.x) — downloaded this pass (gold OA via PMC); verified 8 pp.
Harley 2011 (10.1089/rej.2010.1085) — downloaded this pass; verified 6 pp.

CRITICAL: DOI error found and corrected — the wiki cited 10.1182/bloodadvances.2018018093 for Khincha 2018, which is a completely different paper (Wölfl et al. blinatumomab lineage switch case report). Correct DOI: 10.1182/bloodadvances.2018016964. Archive had downloaded and stored the wrong paper.

Corrections made:

Khincha 2018 DOI corrected: 10.1182/bloodadvances.2018018093 (Wölfl et al.) → 10.1182/bloodadvances.2018016964 (Khincha et al.)
Khincha 2018 journal coordinates corrected: 2(12):1472–1476 → 2(11):1243–1249
Khincha 2018 n added: 10 androgen-treated vs 16 untreated DC patients (total 26); precise stats added: P=.71 unadjusted, P=.24 GEE-adjusted
Townsley 2016 stopping body corrected: “stopped early by the DSMB” → “stopped early by the NHLBI institutional review board” (per paper text)
Townsley 2016 hematologic response corrected: “79–83%” conflation → per-timepoint values (79% at 3 mo, 81% at 6 mo, 78% at 12 mo, 83% at 24 mo); adverse effects (liver-enzyme 41%, muscle cramps 33%) added
Townsley 2016 ITT rate added: 12/27 (44%; 95% CI 26–64; P<0.001) from abstract
Bernardes de Jesus 2011 in-vivo telomere claim corrected: “mean telomere length modestly increased” → average TL NOT significantly increased in the in-vivo arm; it was the percentage of critically short telomeres (below 2–4 kb) that was significantly reduced; DNA damage foci reduction was the in-vitro MEF arm, not in-vivo blood leukocytes; dose 25 mg/kg/day added
Harley 2011 n confirmed from full paper: baseline n=114 (63±12 years, 72% male); per-timepoint ns added

Schema note: clinical-stage: phase-2 on telomerase-activators page vs phase-2-trials on senolytics.md — inconsistency flagged for next CLAUDE.md cleanup pass.

Downstream pages that may need updates:

telomere-attrition — check for any Khincha 2018 citations using wrong DOI
danazol (stub, not yet seeded) — when seeded, use corrected Townsley 2016 numbers
replicative-senescence — already verified; Bodnar numbers confirmed consistent, no update needed

[2026-05-07] verify | molecules/proteins/tpp1.md

Pages verified: 1

molecules/proteins/tpp1.md — corrections: 7 substantive (see below); 4 PDFs downloaded this pass (Nandakumar 2012, Schmidt 2014, Kocak 2014, Latrick 2010)

PDFs read: Wang 2007 (local), Xin 2007 (local), Nandakumar 2012 (downloaded), Schmidt 2014 (downloaded, pp. 1–10 of 20), Kocak 2014 (downloaded, all 13 pp.), Latrick 2010 (downloaded, all 10 pp.). Liu 2004 not_oa — unverified.

Corrections made:

K170Δ mouse model claim REMOVED (seeder fabrication confirmed by PDF): Kocak 2014 is a human genetic study with HeLa cell-based validation only — no knock-in mouse.
TEL patch residue list expanded: “E168, E169, L212” → 7 residues: E168, E169, E171, R180, L183, L212, E215 (Nandakumar 2012 PDF).
Processivity claim corrected: “~6 nt to ~60 nt” (unsourced ~10-fold) → “2–4 fold” with actual R1/2 values from Wang 2007 and Latrick 2010.
Schmidt 2014 charge-swap details added: TERT K78 ↔ TPP1 E215 direct contact (E215K rescues K78E, 25%→98% RAP, p<0.01).
TIN2-binding domain boundary corrected: “~337–406” → “~337–544 (C-terminal)” — P491 is in this domain per Kocak 2014.
OB-fold N-terminal boundary corrected: “~1–260” → “~87–250” per Wang 2007 Fig 1a (first 87 residues dispensable).
All 4 footnotes updated with precise quantitative data and “PDF verified” status.

Downstream pages that may need updates:

tert — if TEL-patch processivity is stated as “~60 nt” or TEL-patch residue list is incomplete
telomere-attrition — if K170Δ mouse is referenced from Kocak 2014
cellular-senescence — check for same K170Δ attribution
shelterin — verify TEL-patch residue list matches corrected 7-residue set

[2026-05-07] verify | molecules/proteins/dkc1.md

Pages verified: 1

molecules/proteins/dkc1.md — corrections: 7 (see below)

PDFs read:

Vulliamy 2001 (10.1038/35096585) — local; verified end-to-end (4 pp.)
Ruggero 2003 (10.1126/science.1079447) — local; verified end-to-end (5 pp.)
Garus & Autexier 2021 (10.1261/rna.078953.121) — downloaded this pass (PMC8594475); verified end-to-end (18 pp.)
Mitchell 1999 (10.1038/990141) — not_oa; tagged no-fulltext-access
Heiss 1998 (10.1038/ng0598-32) — not_oa; tagged no-fulltext-access

External checks:

UniProt O60832 — length 514 aa, PUA 296–371, NLS 2–21 / 446–514 confirmed via REST
NCBI Gene 1736 — DKC1, Xq28 (NC_000023.11) confirmed via efetch
HGNC 2890 — confirmed via UniProt cross-reference
UniProt Q9NX24 (NHP2), Q9NPE3 (NOP10), Q9NY12 (GAR1) confirmed correct
Q9Y224 identified as RTRAF_HUMAN (wrong) — corrected to Q9NY12 for GAR1
GenAge — DKC1 confirmed absent from allgenes.php; genage-id: null is correct

Corrections made:

GAR1 complex-subunit accession: Q9Y224 (RTRAF_HUMAN) → Q9NY12 (GAR1_HUMAN)
Ruggero 2003 pseudouridylation framing corrected: “partially preserved” → “impaired in G1/G2” (10–40% reduction; paper’s primary claim is that rRNA pseudouridylation defects initiate DC)
TERC reduction 50–90% figure gap-tagged as unverifiable (#gap/no-fulltext-access); Mitchell 1999 is closed-access
Vulliamy 2001 mechanism: haploinsufficiency vs dominant-negative correctly qualified per-family per the source
Domain description: “N-terminal coiled-coil” → “N-terminal extension (NTE)” per Garus 2021; added C-terminal NLS/NoLS residue range 446–514
GenAge: “Not found, should be checked” → “Confirmed absent” with verification method
Ruggero 2003 footnote: quantitative parameters added (n=50/16, Hb values, 50% tumor incidence, Ψp/Up % reduction)

Pages unverifiable (closed-access):

Mitchell 1999 (10.1038/990141) — no-fulltext-access
Heiss 1998 (10.1038/ng0598-32) — no-fulltext-access

Downstream pages with independent errors flagged (not propagated here — for main agent):

molecules/proteins/terc.md complex-subunits field: NOP10 listed as P57077 (M3KCL_HUMAN, wrong) and GAR1 as Q9Y285 (FARSA_HUMAN, wrong) — independent errors on that page, not introduced from dkc1.md

[2026-05-07] verify | molecules/proteins/tin2.md

Pages verified: 1

molecules/proteins/tin2.md — corrections: 6 substantive (see below); all 4 non-local PDFs downloaded this pass

PDFs read:

Kim 1999 (10.1038/70508) — local; verified end-to-end (8 pp.)
Ye 2004 (10.1074/jbc.M409047200) — downloaded this pass; verified end-to-end (8 pp.)
Houghtaling 2004 (10.1016/j.cub.2004.08.052) — downloaded this pass; verified end-to-end (11 pp.)
Savage 2008 (10.1016/j.ajhg.2007.10.004) — downloaded this pass; verified end-to-end (9 pp.)
Walne 2008 (10.1182/blood-2008-05-153445) — downloaded this pass; verified end-to-end (7 pp.)

External checks:

UniProt Q9BSI4 REST API — confirmed 451 aa canonical (TIN2L, isoform 1); TIN2S (isoform 2) lacks residues 355–451; PTMs include ADP-ribosylation at E289
NCBI Gene 26277 — confirmed (gene ID matches wiki)
GenAge — confirmed null (no TINF2 entry); wiki’s genage-id: null is correct
HGNC 11824 — portal did not load; not independently re-verified (flagged in verified-scope for next lint pass)

Corrections made:

Kim 1999 — telomere-length regulation direction inverted: “TIN2 overexpression elongates telomeres” → wild-type TIN2 is a NEGATIVE regulator (overexpression maintains/shortens); only N-terminal deletion mutants (TIN2-12, TIN2-13) elongate telomeres in a telomerase-dependent, dominant-negative manner.
Isoform names swapped: TIN2S was listed as the 451 aa canonical form and TIN2L as the shorter form — corrected to match UniProt (TIN2L = 451 aa isoform 1; TIN2S = shorter isoform 2 lacking residues 355–451).
Ye 2004 mechanism: “cells expressing a TIN2 mutant that cannot contact TRF2” → siRNA depletion of TIN2 (not a contact-deficient mutant); TRF2/hRap1 telomeric signal lost in >90% of knockdown nuclei without reducing protein levels.
Savage 2008 n fabricated: “n=20 individuals across 15 families” → Family A (~10 mutation carriers) + 4 unrelated probands + 298 controls; only 1 multi-generation family + 4 de novo cases.
Mutation cluster boundaries and unsupported mutations: “residues 280–300, including R270H/R270C” → 18-aa segment residues 280–298 per Walne 2008; R270H/R270C found in neither source and removed; R282C (Walne 2008) added.
Walne 2008 n wrong: “n=85 DC/bone-marrow-failure patients” → 175 uncharacterized DC patients + 244 related patients screened; TINF2 mutations in 33/175 (18.9%) uncharacterized DC = ~11% of all DC.

Unsourced claims (unchanged, gap-tagged):

GTEx aging correlation — not queried; tagged not-queried
MR causal evidence — tagged not-tested

Downstream pages to check (main agent task):

Any shelterin-complex page citing TIN2 mutation cluster boundaries or isoform description
hallmarks/telomere-attrition.md — may cite TINF2/DC mutation data
Any DC-specific page if it exists (dyskeratosis-congenita.md or similar)

[2026-05-07] verify | molecules/proteins/trf2.md

Pages verified: 1

molecules/proteins/trf2.md — corrections: 7 substantive (see below); Doksani 2013 and Wu 2012 downloaded this pass

PDFs read:

Broccoli 1997 (10.1038/ng1097-231) — local; verified end-to-end (5 pp.)
van Steensel 1998 (10.1016/s0092-8674(00)80932-0) — local; verified end-to-end (13 pp.)
Celli & de Lange 2005 (10.1038/ncb1275) — local; verified end-to-end (7 pp.; previously verified on shelterin page 2026-05-05)
Doksani 2013 (10.1016/j.cell.2013.09.048) — downloaded this pass; verified end-to-end (12 pp.)
Wu/Takai/de Lange 2012 (10.1016/j.cell.2012.05.026) — downloaded this pass; verified end-to-end (8+ pp.)

External checks:

Open Targets Platform ENSG00000132604 — API functional (HTTP 400 seeding error was transient); tractability entries present but reflect pathway-level associations not TRF2-directed agents; tier-4 retained on aging-context grounds
PubMed search TERF2 + keratinocyte + aging — 1 result (PMID 41027193); arsenic-exposure skin damage paper, not the age-expression correlation described; claim remains unsourced
PubMed search Terf2 heterozygous + lymphoma / haploinsufficiency cancer — 0 results; claim remains unsourced

Corrections made:

Imaging modality (Doksani 2013): “super-resolution structured illumination microscopy (SIM)” → STORM (stochastic optical reconstruction microscopy). The paper explicitly uses STORM throughout; SIM is not mentioned.
Cell type and experimental design (Doksani 2013): “human HeLa cells with dominant-negative TRF2” → SV40LT-immortalized mouse MEFs with conditional TRF2 deletion (TRF2^F/- Cre-ER^T1). t-loops reduced 6-fold (36%→6%; p=0.01); other shelterin deletions had no effect. “Restored by full-length TRF2 reconstitution” claim removed — paper does not describe reconstitution.
Cell type (van Steensel 1998): “primary human fibroblasts and HeLa cells” → HTC75 cells (tetracycline-inducible HT1080 fibrosarcoma derivative). Primary fibroblasts appear only in discussion as a comparator. Senescence-like phenotype was demonstrated in HTC75, not primary fibroblasts.
Fusion phenotype quantification (van Steensel 1998): added ≥10-fold increase, 88% metaphases with at least one fusion at day 6; removed unsupported “extrachromosomal telomeric circles” label (paper describes enlarged TTAGGG restriction fragments, not circles).
ATM timing (Celli 2005): “within 24–48 h” → 48–72 h post-Cre (matches paper text and Fig 3c/4c).
Apollo/overhang generation nuance (Wu 2012): corrected to specify Apollo affects leading-strand telomeres specifically (30-40% reduction); Exo1 contributes at both ends independently of TRF2; POT1b-CST corrects overhangs in G1. The exclusive TRF2-Apollo claim was overstated.
Domain residue boundaries (Broccoli 1997): clarified that precise residue numbers in the domain table are from post-1997 structural work; Broccoli 1997 covers domain identification and homodimerization only.
Druggability note: Open Targets API error corrected; tier-4 reasoning updated to reflect successful API query.
K5TRF2 skin cancer: corrected from “without evidence of telomere shortening protection benefit” to accurate description (TRF2 overexpression causes critically short telomeres and genomic instability via deregulated XPF/ERCC1; citations Blanco 2007, Stout & Blasco 2009 added).

Unsourced claims (unchanged, still unsourced):

Age-related TRF2 expression decline in skin keratinocytes — no primary source found in PubMed search
Terf2+/- haploinsufficiency lymphoma phenotype — no primary source found

Downstream pages to check (main agent task):

molecules/proteins/shelterin.md — cites Celli 2005 with 119±18% overhang figure (confirmed correct); no correction needed
hallmarks/telomere-attrition.md — may cite Celli 2005 or Doksani 2013 claims; check for SIM/STORM error propagation
Any page citing van Steensel 1998 claims about primary fibroblasts should be checked for cell-type accuracy

[2026-05-07] verify | molecules/proteins/pot1.md

Pages verified: 1

molecules/proteins/pot1.md — corrections: 5 (see below); Robles-Espinoza 2014 downloaded this pass

PDFs read:

Hockemeyer 2006 (10.1016/j.cell.2006.04.044) — local; verified end-to-end (15 pp.)
Wang 2007 (10.1038/nature05454) — local; verified end-to-end (5 pp.)
Robles-Espinoza 2014 (10.1038/ng.2947) — downloaded this pass (PMC OA); verified end-to-end (5 pp.)
Baumann 2001 (10.1126/science.1060036) — not_oa; unverifiable; tagged no-fulltext-access
Ramsay 2013 (10.1038/ng.2584) — download failed (two attempts); confirmed as CLL paper via Crossref; no-fulltext-access

External checks:

UniProt Q9NUX5 — length 634 confirmed via REST API
NCBI Gene 25913 — confirmed POT1, Homo sapiens
Open Targets ENSG00000128513 — tier-4 confirmed (no SM/AB/PR approved drugs; only proteomics metadata)
GenAge human_genes CSV — POT1 not present; genage-id: null confirmed correct

Corrections made:

Domain table overhauled: five-row table (OB1/OB2/Linker/C-OB3/HJR-like with unsupported residue ranges) replaced with two-row bipartite table matching Wang 2007 Fig 1a (POT1-N ~1–300; POT1-C ~300–634) and Hockemeyer 2006 Fig 1A (Telomere/TPP1 binding begins at ~415 within POT1-C). Unsupported “HJR-like domain” label removed; sub-domain detail redirected to Lei 2004 with gap tag.
Pot1b null phenotype: “shortened G-overhangs” → “elongated G-overhangs” (7–11x increase in overhang signal in liver, kidney, spleen per Hockemeyer 2006 Fig 6C). This was a direction-reversal error.
TEL-patch claim: decoupled from Wang 2007 citation. Wang 2007 does not use the term “TEL patch” — that term is from Nandakumar et al. 2012 (Science). Section revised to describe what Wang 2007 actually shows (structural homology to TEBPβ, processivity quantification). Gap tag added.
ATR suppression citation: changed from [^baumann2001] to [^hockemeyer2006] for the Pot1a-specific TIF phenotype claim; Baumann 2001 is the POT1 discovery paper (S. pombe + human cell extracts) and does not establish the ATR-RPA mechanistic connection.
Pot1a phenotype qualifier: “massive chromosomal instability” label qualified — Hockemeyer 2006 shows Pot1a single KO causes TIF formation in ~30% of cells; the 70–80% γH2AX phenotype (widespread DDR) is the DKO (Pot1a + Pot1b double KO) phenotype. Table note updated accordingly.

Additional fixes:

mr-causal-evidence: not-tested → partial (GWAS instruments exist at POT1 region; MR study pending)
Robles 2014 footnote: added n=184 patients from 105 pedigrees, 4/105 families with POT1 variants, specific p-values and variant names (Tyr89Cys, Gln94Glu, Arg273Leu)
Wang 2007 footnote: added processivity quantification (R₁/₂ 0.78→3.3); removed “TEL-patch contacts TERT” language
Ramsay 2013 footnote: added no-fulltext-access noting PDF download failure; abstract/title-only verification

Unverifiable claims (not_oa or download-failed sources):

Baumann 2001 OB-fold ssDNA binding characterization: tagged no-fulltext-access in footnote
Ramsay 2013 CLL cohort size, somatic mutation frequency: tagged no-fulltext-access in footnote

Downstream pages to check (main agent task):

molecules/proteins/shelterin.md — cites POT1 length (634 aa, confirmed unchanged); shelterin page already correctly shows elongated-overhang Pot1b phenotype context; no correction needed
hallmarks/telomere-attrition.md — does not directly cite Hockemeyer 2006 or the Pot1a/b phenotype; no correction needed

[2026-05-07] verify | processes/replicative-senescence.md

Pages verified: 1

processes/replicative-senescence.md — corrections: 5 (see below); Herbig 2004 downloaded this pass

PDFs read:

d’Adda di Fagagna 2003 (10.1038/nature02118) — local; verified end-to-end
Bodnar 1998 (10.1126/science.279.5349.349) — local; verified end-to-end
Coppé 2008 (10.1371/journal.pbio.0060301) — local; verified end-to-end
Herbig 2004 (10.1016/s1097-2765(04)00256-4) — downloaded this pass (bronze OA via Cell camoufox); verified end-to-end
Hayflick 1961 (10.1016/0014-4827(61)90192-6) — not_oa; unverifiable
Harley 1990 (10.1038/345458a0) — not_oa; unverifiable
Campisi 2007 (10.1038/nrm2233) — not_oa; unverifiable
de Lange 2005 (10.1101/gad.1346005) — download failed; unverifiable

Corrections made:

Bodnar 1998 delivery method: corrected “retroviral vector” to “electroporation with plasmid expression constructs (MPSV-hTRT or SV40-hTRT vectors)” — paper used electroporation, not retroviral delivery.
Bodnar 1998 PD numbers: expanded from “>20 PDs” to the two-cell-type breakdown: RPE ~20 PD (P < 10^-24) and BJ 36 PD (P < 10^-6) beyond respective controls. Added n per group in footnote.
Bodnar 1998 “safety caveat”: removed incorrect claim that TERT-transduced cells “eventually arrested at a subsequent limit” — the paper explicitly states this was not determined in the study duration; cells continued to divide rapidly as of galley proofs. Tagged unsourced.
Checkpoint-rescue attribution: corrected attribution in Step 2 narrative — the kinase-dead construct + BrdU 17% experiment is d’Adda di Fagagna 2003, not Herbig 2004. Herbig 2004’s siRNA result is specific to p16-negative cells only (rescued ~60% BrdU+); p16-positive cells are NOT rescued by ATM/ATR ablation — this critical nuance was absent.
SASP paragraph: removed incorrect attribution of C/EBPβ and cGAS-STING regulation to Coppé 2008 — that paper characterizes the secretory phenotype across senescence types (NF-κB-driven) but does not characterize C/EBPβ as a SASP driver (that is Kuilman 2008) nor cGAS-STING (that is Glück 2017). Sentence restructured to cite Coppé 2008 only for findings it contains, and notes that additional regulators characterized in subsequent work.
Coppé 2008 SASP comparison: improved specificity — replaced “largely overlapping but distinct” with the actual quantitative correlation data: REP vs XRA r > 0.9; RAS-induced SASP slope > 1 vs REP/XRA; 5 proteins uniquely elevated in RAS SASP. Footnote updated with paper’s actual methodology and key quantitative results.
Banner removed (was auto-extraction warning).

Unverifiable claims (not_oa or download-failed sources):

Hayflick 1961 PD counts (~40–60), cell types (WI-38, MRC-5): tagged no-fulltext-access in footnote; numbers consistent with d’Adda di Fagagna 2003 citing PD ~45 for senescent MRC5
Harley 1990 shortening rate (~50–150 bp/division): tagged no-fulltext-access in footnote
Campisi 2007 baboon TIF claim and TERT-KO G4–G5 details: tagged no-fulltext-access; noted Jeyapalan et al. 2007 Mech Ageing Dev as likely primary source for baboon TIF in vivo
de Lange 2005 shelterin subunit list: tagged no-fulltext-access; subunit list consistent with d’Adda di Fagagna 2003 citing TRF1/TRF2

Cross-check with verified TERT page:

Bodnar 1998 numbers on TERT page (verified 2026-05-05): RPE ~20 PD; BJ 36 PD; P < 10^-24 and P < 10^-6 — consistent. No divergence.

Cross-check with verified SASP page:

Coppé 2008 n=5 fibroblast strains confirmed on SASP page — consistent.
SASP page correctly attributes C/EBPβ to Kuilman 2008 and cGAS-STING to Glück 2017 — replicative-senescence page now matches this.

Downstream pages that may need updates (main agent to handle):

hallmarks/cellular-senescence.md — may repeat the “retroviral vector” Bodnar claim; check
hallmarks/telomere-attrition.md — may cite Bodnar 1998 with retroviral vector language; check
Any page citing Herbig 2004 that asserts ATM siRNA rescues arrest without the p16-negative qualifier

[2026-05-07] verify | molecules/proteins/terc.md

Pages verified: 1

molecules/proteins/terc.md — corrections: 5 (see below)

PDFs read:

Vulliamy 2001 (10.1038/35096585) — local; verified end-to-end
Armanios 2007 (10.1056/NEJMoa066157) — local; verified end-to-end
Theimer 2003 (10.1073/pnas.242720799) — downloaded this pass (green OA via PMC)
Blasco 1997 (10.1016/s0092-8674(01)80006-4) — downloaded this pass (bronze OA via Cell camoufox)
Theimer 2005 (10.1016/j.molcel.2005.01.017) — downloaded this pass (bronze OA via Cell camoufox)
Leeper 2005 (10.1261/rna.7222505) — downloaded this pass (hybrid OA via RNA journal)
Feng 1995 and Mitchell 1999 — closed-access (not_oa); unverifiable

Corrections made:

Blasco 1997 body text: removed incorrectly-attributed in vivo tissue phenotypes (testicular atrophy, bone marrow failure, gut atrophy) — these were from Lee 1998 and Rudolph 1999, not Blasco 1997. Replaced with what Blasco 1997 actually showed: telomere shortening at 4.8±2.4 kb/generation, cytogenetic abnormalities from G4+, and cell/tumor biology.
Theimer 2003 footnote: corrected description — paper uses pseudoknot–hairpin equilibrium framing and stem 2 / j2b3 loop terminology (not “P-loop and J2a/2b junctions”); paper characterizes structural thermodynamics, not direct telomerase activity measurements.
Theimer 2005 footnote: updated to reflect what the PDF confirms (triple-helix network at helical junction; not CR4/CR5); removed “download pending” status.
Armanios 2007 mechanism statement: corrected from “senescence → fibroblast activation → fibrosis” to paper’s actual proposal (cell loss → remodeling response); added specific p-values and n from the paper.
Blasco 1997 and Leeper 2005 footnotes: updated archive status from “pending” to “local PDF available.”

Identity fields verified:

NCBI Gene 7012: confirmed TERC, Homo sapiens, 3q26.2 ✓
HGNC 11727: confirmed TERC, locus type = RNA long non-coding, Ensembl ENSG00000270141 ✓
is-noncoding-rna: true ✓
uniprot: null with no-uniprot-noncoding-rna ✓

Citation corrections confirmed correct:

Theimer 2003 (PNAS) replacing “Wong 2003 PNAS” (nonexistent): confirmed correct
Leeper & Varani 2005 RNA for CR4/CR5 structure: confirmed correct against PDF
Theimer 2005 Mol Cell for pseudoknot structure: confirmed correct against PDF

Pages unverifiable (closed-access):

Feng 1995 (10.1126/science.7544491) — not_oa; tagged no-fulltext-access
Mitchell 1999 (10.1038/990141) — not_oa; tagged no-fulltext-access

Downstream pages that may need updates (main agent to handle):

hallmarks/telomere-attrition.md — may cite Blasco 1997 for in vivo tissue phenotypes; check attribution
studies/blasco-1997-terc-knockout.md if it exists — verify it doesn’t overattribute tissue phenotypes to Blasco 1997

Schema escalation flagged:

druggability-tier: 1 for ncRNA target (imetelstat/TERC) has no CLAUDE.md guidance; flagged for schema cleanup pass

[2026-05-07] verify | tak1

Pages verified: 1 (partial — Wang 2001 PDF read end-to-end; UniProt REST-verified; 5 other DOIs deferred pending PDF access)

molecules/proteins/tak1.md

Sources checked:

doi:10.1038/35085597 (Wang 2001, Nature) — read in full (4 pages). TAK1-TAB1-TAB2 (TRIKA2) as minimal IKK-activating unit confirmed. K63-ubiquitin mechanism confirmed. IKKβ Ser177/Ser181 and MKK6 S207/T211 phosphorylation confirmed. TAB3 not described in this paper — double-KO attribution removed.
UniProt O43318 — REST-verified: 606 aa canonical, kinase domain 36–291, active site Lys156, Thr187/Ser192/Thr184 autophosphorylation confirmed.
doi:10.1074/jbc.M207453200 (Ninomiya-Tsuji 2003) — confirmed via Crossref; correct paper, DOI was already correct.
doi:10.1038/ncb2784 (Acosta 2013) — confirmed via Crossref; covers paracrine senescence via inflammasome, NOT the IL-1α autocrine loop. SASP autocrine claim reattributed to Orjalo 2009 PNAS (doi:10.1073/pnas.0905299106).
doi:10.1161/CIRCULATIONAHA.114.011195 (Li et al. 2014, Circulation) — confirmed as correct cardiac TAK1 cKO paper. Seeder’s “Liu 2013 Cell Death Dis / doi:10.1038/cddis.2013.341” was wrong (resolves to optic nerve astrocyte paper unrelated to TAK1).
Open Targets API (ENSG00000135341) — druggability tier 2 confirmed.

Corrections applied:

Removed non-schema caused-by-note: field; moved content to new body subsection
TAB2/TAB3 double-KO claim removed from [^wang2001] attribution
SASP autocrine loop: added [^orjalo2009] (Orjalo PNAS 2009) as primary source; added [^acosta2013] correctly as paracrine senescence
Cardiac cKO: added [^li2014] (Li 2014 Circulation) replacing the gap placeholder
Removed ⚠️ auto-extraction banner

Deferred (PDF unavailable): Sato 2005 (not_oa), Wade 2016 (pending), Yuan 2023, Wang 2025, Totzke 2017 (pending)

Downstream pages to check (main agent to propagate):

il-1-signaling — verify TAB3 double-KO not cited from Wang 2001
traf6 — may inherit SASP autocrine claim

[2026-05-07] verify | myd88

Pages verified: 1 (partial — 4 of 9 cited PDFs verified end-to-end; UniProt REST-verified; 4 deferred pending PDF access)

molecules/proteins/myd88.md

Sources checked:

doi:10.1038/nature09121 (Lin 2010, Nature) — read in full (7 pages). One correction applied: Myddosome layer order reversed (MyD88 is basal, not apical; IRAK2 is apical, not basal).
doi:10.1016/s1074-7613(00)80402-1 (Wesche 1997, Immunity) — read in full (11 pages). No corrections; wording consistent with verified il-1-signaling page.
doi:10.1126/science.1158298 (von Bernuth 2008, Science) — read in full (7 pages). No corrections; n=9, 5 kindreds, age-amelioration mechanism confirmed.
doi:10.1038/nature09671 (Ngo 2011, Nature) — read in full (7 pages). No corrections; L265P in 29% ABC-DLBCL, n=382 biopsies confirmed.
UniProt Q99836 — verified via REST API. All identity fields confirmed (length 296, DD 54-109, ID 110-155, TIR 159-293).
doi:10.1056/NEJMoa1707914 (Ridker 2017, CANTOS) — cite-by-context only; confirmed consistent with canakinumab.md canonical home.

Deferred (PDF unavailable):

doi:10.1074/jbc.m109.022392 (Motshwene 2009) — archive status: pending
doi:10.1056/nejmoa1200710 (Treon 2012) — archive status: pending
doi:10.1189/jlb.1206746 (Loiarro 2007) — archive status: not_oa
doi:10.1182/blood.2021011570 (Kovtonyuk 2022) — archive status: pending

Corrections made: 1

myd88.md § Myddosome assembly: “MyD88 DDs form the apical platform; IRAK4 DDs occupy the middle; IRAK2 DDs form the base” → “MyD88 DDs form the basal platform (bottom two layers); IRAK4 DDs occupy the middle layer; IRAK2 DDs form the apical layer at the top”

Downstream pages to check: none — no other page repeats this exact layer-order claim. The il-1-signaling page does not describe Myddosome layer order.

[2026-05-07] verify | traf6

Pages verified: 1 (partial — 3 of 6 cited PDFs read end-to-end; 1 via PMC full text; 1 closed access; 1 not downloaded)

molecules/proteins/traf6.md

Sources checked:

doi:10.1016/s0092-8674(00)00126-4 (Deng 2000, Cell) — local PDF read in full (5 pages). K63 chain mechanism, Ubc13/Uev1A (TRIKA1) requirement confirmed. No corrections needed to mechanistic claims.
doi:10.1101/gad.13.8.1015 (Lomaga 1999, Genes & Dev) — local PDF read in full (10 pages). Corrections applied: osteoclast characterisation corrected (present but non-functional, not absent), signaling defect wording corrected (IL-1 NF-κB impaired not abolished; JNK absent; T-cell responses not significantly different from WT). Strain background added (129J → C57BL/6), survival data quantified (11% at 2 wk).
doi:10.1074/jbc.RA118.002649 (Brenke 2018, JBC) — local PDF read in full (6 pages). Corrections applied: cIAP1 cross-reactivity added to selectivity description; in vivo doses added (psoriasis ~1.5 mg/kg topical, n=8/group; RA 6/10/14 mg/kg IP ×14 days). C25-140 binding to TRAF6 directly (not Ubc13) confirmed by NMR.
doi:10.1016/j.exger.2014.01.007 (Hinojosa 2014, Exp Gerontol) — PMC3989429 full text read. A20 elevation in aged macrophages confirmed. Strain (C57BL/6), ages (4/12/21 mo), organism (S. pneumoniae TIGR4), fish oil effect (~100-fold bacterial reduction) verified and added to footnote. Local download failed (green OA, archive status: failed).
UniProt Q9Y4K3 — REST API: all 5 domain positions confirmed (RING 70-109, TRAF-type ZF1 150-202, ZF2 203-259, coiled-coil 288-348, MATH 350-499), sequence length 522 aa, all PTM sites confirmed.
NCBI Gene 7189, HGNC 12036, Ensembl ENSG00000175104 — confirmed via APIs.
Open Targets ENSG00000175104 — API retry successful (no 500 error). Tractability: no SM/AB/PR approved drug or clinical compound; druggability tier 3 confirmed. Gap annotation #gap/opentargets-api-error removed.

Corrections applied:

Lomaga 1999 osteoclast phenotype: “absent functional osteoclasts” → “osteoclasts present in normal numbers but non-functional (withdrawn from bone surface, lacking ruffled borders)”
Lomaga 1999 IL-1 signalling: “abolished” → “impaired (NF-κB) / absent (JNK/SAPK)”
Lomaga 1999 T-cell description: “T-cell development intact” → “T-cell proliferative responses not significantly different from WT”
Lomaga 1999 survival: added 11% pup survival at 2 wk, died prematurely within weeks
Lomaga 1999 strain: added 129J ES cells → C57BL/6 background
Brenke 2018 selectivity: added cIAP1 cross-reactivity; added RA dose 6/10/14 mg/kg IP and psoriasis dose ~1.5 mg/kg topical; added n=8/group
Brenke 2018 mechanism: added NMR confirmation of direct TRAF6 binding; added that Ubc13–Uev1A interaction is unaffected
Open Targets: removed stale error annotation; added API-verified tractability summary
Hinojosa 2014: added strain/ages/organism/effect-size to footnote
Banner removed; verified flag flipped

Deferred (PDF unavailable):

doi:10.1046/j.1365-2443.1999.00265.x (Naito 1999) — not in PMC; archive status: pending
doi:10.1016/s0014-5793(04)00505-8 (FEBS Lett 2004) — closed access (not_oa)
doi:10.1016/j.jmb.2021.166844 (Das 2021) — archive not checked

Downstream pages to check (main agent to propagate):

il-1-signaling — check that osteoclast/Lomaga claim is not echoed there
chronic-inflammation — TRAF6 mechanistic role description may inherit any of the corrected claims

[2026-05-07] verify | hsp90

Pages verified: 1 (partial — 4 of 6 cited PDFs verified; 2 not_oa)

molecules/proteins/hsp90.md

Sources checked:

doi:10.1073/pnas.91.18.8324 (Whitesell 1994, PNAS) — downloaded and read in full (5 pages). Corrections applied.
doi:10.1016/j.cell.2012.06.047 (Taipale 2012, Cell) — downloaded and read in full (15 pages). Multiple corrections applied.
doi:10.1016/j.cell.2015.03.032 (Walther 2015, Cell) — downloaded and read in full (14 pages). Corrections applied.
doi:10.1101/cshperspect.a034017 (Biebl & Buchner 2019, Cold Spring Harb Perspect Biol) — downloaded and read (10 pages). Author name correction applied.
doi:10.1038/24550 (Rutherford & Lindquist 1998, Nature) — not_oa; qualitative capacitor claims retained with gap tags.
doi:10.1038/s41580-019-0101-y (Hipp 2019, Nat Rev Mol Cell Biol) — not_oa; qualitative proteostasis network claims retained with gap tags.
UniProt P07900, NCBI Gene 3320, HGNC 5253, Ensembl ENSG00000080824 — all verified via live REST APIs 2026-05-07. All correct.
Open Targets Platform ENSG00000080824 — SM tractability queried: Advanced Clinical TRUE, Approved Drug FALSE.

Corrections made to molecules/proteins/hsp90.md:

druggability-tier: 1 → 2 (Open Targets confirmed no approved drug exists; advanced clinical probes only; frontmatter and body updated)
Taipale 2012 paper title corrected: was “HSP90 at the hub of protein homeostasis: emerging mechanistic insights” (that is a different Taipale 2010 Nat Rev paper); correct title is “Quantitative Analysis of Hsp90-Client Interactions Reveals Principles of Substrate Recognition”
Taipale 2012 study design corrected: was “proteomics / quantitative IP-MS after HSP90 crosslinking”; correct: “original research — LUMIER/BACON in-cell luminescence interaction assay”
Taipale 2012 client count: “~200 high-confidence clients” → “almost 400 client proteins” (paper’s discussion wording); “~10% of human proteome” removed (not stated in paper)
TF interaction rate: “~30% of TFs” → “~7% of tested TFs (58/843 unique TFs)” — paper explicitly states only ~7% of TFs interact with HSP90, far lower than the previous claim
Kinase rate table: “~60% of kinome” qualified to “61% of tested kinase clones (193/420)” with explicit coverage note
Whitesell 1994 mechanism: removed false NTD ATP-pocket attribution; paper identified geldanamycin→HSP90 binding but did not characterize the binding site; added Stebbins 1997 note + needs-citation
Whitesell 1994 cell model: expanded from “NIH 3T3 cells” to include PC-3M, Jurkat, CHP-100 and reticulocyte lysate (all used in paper)
Walther 2015 protein count: “~700 proteins” → “975 of 1,083 proteins” (no ~700 figure in paper; 975 is the reported count at day 12)
Walther 2015 mechanism reframed: supersaturation/abundance is the primary aggregate driver per paper; small HSPs aggregate as a protective response; “chaperone network saturation” framing overstates the paper’s conclusion
Walther 2015 aggregation timing: “beginning at reproductive maturity” → “mainly after day 6 of adulthood (post-reproductive)” per paper text
Biebl & Buchner 2019 author list corrected: “Schopf FH, Biebl MM & Buchner J” → “Biebl MM & Buchner J” (two authors; Schopf FH was not an author on this paper)

Pages unverifiable (not_oa, claims retained with gap tags):

Rutherford & Lindquist 1998 (Nature 24550) — evolutionary capacitor concept; no-fulltext-access
Hipp 2019 (Nat Rev Mol Cell Biol) — proteostasis network framework; no-fulltext-access

Downstream pages that may need review (do not propagate here — for main agent):

pathways/heat-shock-response.md — may cite HSP90 druggability-tier-1; update to tier-2
Any page citing Taipale 2012 for “~30% of TFs” or “~60% of kinome” — both figures require qualification

[2026-05-07] verify | il-1-signaling + nf-kb cross-fix

Pages verified: 1 (pathways/il-1-signaling.md); 1 cross-page correction (pathways/nf-kb.md)

Sources checked (il-1-signaling.md):

doi:10.1056/NEJMoa1707914 (Ridker 2017 CANTOS, NEJM) — local PDF at ; read end-to-end (10 pages). All primary endpoint numerics verified exact.
doi:10.1016/s1074-7613(00)80402-1 (Wesche 1997, Immunity) — local PDF confirmed; read end-to-end (6 pages). MyD88 mechanism verified.
doi:10.1074/jbc.270.23.13757 (Greenfeder 1995, JBC) — local PDF confirmed; read end-to-end (6 pages). IL-1RAcP coreceptor identification verified.
doi:10.1182/blood-2010-07-273417 (Dinarello 2011, Blood) — PMID 21304099 confirmed via PubMed eutils; abstract only (not_oa for full text); title and 11-member IL-1 family claim verified via abstract.
doi:10.1182/blood.2021011570 (Kovtonyuk 2022, Blood) — PMID 34525198 confirmed via PubMed eutils; title “IL-1 mediates microbiome-induced inflammaging of hematopoietic stem cells in mice” and mouse model confirmed via abstract.
Reactome R-HSA-9020702: confirmed = “Interleukin-1 signaling” via ContentService API — ID is correct for il-1-signaling.md.
KEGG hsa04060: confirmed as cytokine-cytokine receptor interaction map; not IL-1-specific (noted in page body).

Corrections made to pathways/il-1-signaling.md:

Fatal infection signal: “HR ~1.7 vs placebo” → “incidence rate ratio ~1.7 (0.31 vs 0.18 events per 100 person-years; P=0.02)” in both body text and footnote. The paper (Table 3) reports incidence rates, not a formal Cox HR, for this endpoint.
Per-arm n added to Ridker 2017 footnote: placebo n=3344; 50 mg n=2170; 150 mg n=2284; 300 mg n=2263.
MyD88 IRAK recruitment mechanism: “MyD88 DEATH domain recruits IRAK4” → “MyD88 mediates IRAK recruitment to the receptor complex.” Wesche 1997 shows IRAK interaction is mediated by the full MyD88 molecule binding to the IL-1R1/IL-1RAcP heterocomplex; the N-terminal death domain activates NF-κB independently but the paper does not support attributing IRAK recruitment specifically to the death domain.
Banner removed; verified: true set.

Cross-page correction to pathways/nf-kb.md:

reactome: R-HSA-9020702 → reactome: null with inline note. R-HSA-9020702 is “Interleukin-1 signaling” (confirmed). Reactome has no single top-level NF-κB pathway entry equivalent to KEGG hsa04064; relevant sub-entries listed in comment. KEGG hsa04064 confirmed correct as the primary NF-κB canonical ID. Scope note banner in nf-kb.md updated to reflect the correction.

Downstream pages to check for propagation (for main agent):

canakinumab — contains per-arm CANTOS numerics; check n values and fatal infection terminology against this pass’s corrections
il-1b — may cite Ridker 2017 for CANTOS; check fatal infection language

[2026-05-07] verify | anakinra

Pages verified: 1 (partial — Larsen 2007 PDF verified; 5 other sources not available as local PDFs)

molecules/compounds/anakinra.md

Sources checked:

doi:10.1056/NEJMoa065213 (Larsen 2007, NEJM) — local PDF confirmed and read end-to-end (10 pages). Corrections applied.
ChEMBL CHEMBL1201570 — API confirmed: pref_name ANAKINRA, molecule_type Protein, first_approval 2001, INN anakinra. Molecular formula field null in API (biologic; ChEMBL does not compute formula for proteins).
ClinicalTrials.gov v2 re-queried 2026-05-07: 29 trials RECRUITING or ACTIVE_NOT_RECRUITING (seeder had 3; count updated).
PubChem name lookup for anakinra: 404 (no CID assigned; confirmed null).
Salmon 2022 (doi:10.1016/j.avsg.2022.05.024) — abstract fetched via PubMed. AAA directionality confirmed: IL-1α disruption worsens AAA in mice. Wiki claim correct.

Corrections made to molecules/compounds/anakinra.md:

Per-arm n: “35 anakinra” → “34 anakinra” in footnote (paper Table 1: N=34 vs N=35; 70 randomized total is correct)
HbA1c p-value: “P<0.05” → “P=0.03” with CI 0.01–0.90 (from paper Figure 2A and Results)
Beta-cell endpoints: added proinsulin/insulin ratio P=0.005 and C-peptide AUC P=0.05 specifics
Insulin sensitivity: clarified clamp P=0.58 and HOMA P=0.60 (unchanged, as wiki stated, now with p-values)
Injection-site reactions: added 17/34 (50%) anakinra vs 0/35 placebo to body and footnote (from Table 2)
PK half-life: “~4–6 hours” → “~6–8 hours” in PK table, prose, and comparison table (paper Discussion: “short half-life of anakinra (6 to 8 hours)”)
ClinicalTrials.gov count: 3 → 29 (frontmatter + body updated); 3 most aging-relevant trials noted in table
Verified flag: false → true (partial scope); banner updated

Pages unverifiable (PDFs not in archive):

Bresnihan 1998 — not_oa
Brucato 2016 — pending
Abbate 2022 — pending
Ridker 2016 — pending
Salmon 2022 — pending (abstract only via PubMed)

Downstream pages that may need review:

studies/larsen-2007-anakinra-t2d — if this page exists, per-arm n and p-values need same corrections

[2026-05-07] verify | unfolded-protein-response

Pages verified: 1 (partial)

processes/unfolded-protein-response

Sources checked:

doi:10.1038/nrm3270 (Hetz 2012, Nat Rev Mol Cell Biol) — local PDF confirmed and read end-to-end (14 pages). Three-arm architecture, GRP78/BiP sensor model, PERK/IRE1α/ATF6 mechanism steps, adaptive vs. terminal UPR framework, and cell-type heterogeneity claims all verified.
doi:10.7554/eLife.62048 (Krukowski 2020, eLife) — OA; fetched via web. ISRIB dose (2.5 mg/kg i.p. × 3 days), strain (C57BL/6J), ATF4 reduction, spine density improvement confirmed.
doi:10.1016/j.cell.2013.05.042 (Taylor & Dillin 2013, Cell) — paywalled; not in archive. XBP-1s lifespan claim (~15%) NOT verified against primary source.

Corrections made:

PERK Thr982 autophosphorylation residue removed — Hetz 2012 does not specify this residue; was unsourced.
ATF6β-as-repressor claim softened — Hetz 2012 lists ATF6β as a family member but does not describe it as a repressor; corrected to “functional distinction not fully characterized.”
ISRIB table entry made more precise — added dose, route, strain, ATF4 and spine density findings sourced from Krukowski 2020.
Banner updated from full-warning to partial-verification note specifying which claims remain unverified.
verified: true with verified-scope describing partial coverage.

Unverified claims retained (pending access):

Taylor/Dillin 2013 XBP-1s ~15% lifespan + cell-nonautonomous mechanism (Cell, paywalled)
Brown 2012 GRP78 aging decline (archive pending)
Han/Kaufman 2013 ATF4-CHOP oxidative death (not_oa)
Estebanez 2018 UPR aging review (archive pending)
Mercado 2018 GSK2606414 neuroprotection (archive pending)

Downstream pages to check for propagation: none identified (no entity pages currently cite unfolded-protein-response.md claims with corrected specifics; PERK Thr982 was page-local).

[2026-05-07] verify | waziry-2023 study

Pages verified: 1

studies/waziry-2023-calerie-epigenetic-clock

Source checked:

doi:10.1038/s43587-022-00357-y — local PDF confirmed at (local PDF); read end-to-end (all 10 pages, Nature Aging vol 3, pp 248–257, 2023)

Corrections made:

DNAm array platform: “Illumina EPIC / 450K methylation arrays” → “Illumina Infinium Methylation EPIC BeadChip arrays” — paper specifies EPIC only; 450K was not used.
p<0.005 significance threshold attribution: “pre-registered for this analysis per the authors” → “chosen as a conservative threshold following guidance from Benjamin et al. 2018 (not pre-registered; the clock analysis was post-hoc)” — the paper explicitly states this was a post-hoc analysis and the threshold follows field guidance, not pre-registration.

All key numerics confirmed exact against PDF:

DunedinPACE: d=−0.29 (CI −0.45,−0.13) at 12mo, d=−0.25 (CI −0.41,−0.09) at 24mo, p<0.003 — exact match
PhenoAge: d=−0.03/+0.05, p>0.50 — exact match
GrimAge: d=−0.04/+0.05, p>0.40 — exact match
IV analysis: d=−0.43 (CI −0.67,−0.19) at 12mo, d=−0.40 (CI −0.67,−0.12) at 24mo — exact match
CR achieved: 11.9% ± 0.7% s.e.m. — exact match
n=220 randomized, 197 DNAm analysis sample (128 CR, 69 AL), 185 at 24mo (117 CR, 68 AL) — exact match
Volume/pages: vol 3, pp 248–257 — confirmed correct
“2–3% slowing” language: confirmed present in paper itself (page 249 and 252), not extrapolated from Cohen d alone
Dose-response threshold: >10% vs <10% achieved CR confirmed; 20% used only for IV analysis
Author list: 22 authors confirmed; order and names match frontmatter exactly

Downstream propagation needed (for main agent):

caloric-restriction — cites Waziry 2023; DunedinPACE array platform detail may appear there
dunedinpace-2022 — cites Waziry 2023 as validation; no numeric corrections propagate
phenoage-2018 — cites Waziry null result; no corrections
grimage-2019 — cites Waziry null result; no corrections
epigenetic-alterations — may cite Waziry 2023; no numeric corrections

[2026-05-06] verify | interventions/lifestyle/methionine-restriction.md

Pages verified: 1

Sources checked (9 DOIs):

doi:10.1093/jn/123.2.269 (Orentreich 1993, J Nutr) — DOI lookup failed; verified via PubMed abstract; diet percentages 0.17%/0.86%, ~30% lifespan extension, male Fischer 344 rats confirmed; absolute lifespan values sourced via Fang 2022 citation no-fulltext-access
doi:10.1111/j.1474-9726.2005.00152.x (Miller 2005, Aging Cell) — local PDF (PMC OA), read end-to-end
doi:10.1016/j.exger.2017.01.012 (Ables 2017, Exp Gerontol) — DOI lookup failed; no PMC full-text no-fulltext-access
doi:10.1146/annurev-nutr-062320-111849 (Fang 2022, Annu Rev Nutr) — local PDF (camoufox), read end-to-end (pp. 201–226)
doi:10.1126/sciadv.abn3868 (Tang 2022, Sci Adv) — local PDF (PMC gold OA), read end-to-end
doi:10.3389/fcell.2020.00715 (Kitada 2020, Front Cell Dev Biol) — local PDF (PMC gold OA), read end-to-end
doi:10.1016/j.cmet.2014.02.009 (Solon-Biet 2014, Cell Metab) — local PDF (camoufox), read end-to-end (key pages)
doi:10.1186/s12967-020-02288-x (Olsen 2020, J Transl Med) — local PDF (PMC gold OA), read end-to-end
doi:10.1111/acel.13629 (Plummer & Johnson 2022, Aging Cell) — local PDF (PMC gold OA), read end-to-end

Corrections made:

CRITICAL — Miller 2005 mouse strain corrected: “(C57BL/6J × DBA/2J)F1” → “(BALB/cJ × C57BL/6J)F1 (CB6F1)”. n=40/group confirmed. Diet levels confirmed: control 0.43%, MR starting 0.1%→0.12%→0.15%. Quantitative hormone outcomes added: IGF-I 397→257 ng/mL, insulin 1.6→0.4 ng/mL, glucose 64.7→32.9 mg/dL. Max lifespan 1144 vs ≥1261 days, log-rank p=0.02.
CRITICAL — Olsen 2020 trial duration corrected: “8 weeks” → “7 days”. n corrected: “~50” → “20 randomized / 19 completed (n=6–7/group)”. FGF21 increase quantified: 47% (96→141 pg/mL, p=0.004 group×time). Design clarified: double-blind, 3-arm, isocaloric, vegan-based.
CRITICAL — Ables 2022 author attribution corrected: “(Ables et al.)” → “Plummer JD, Johnson JE”. Study details added (n=4/group male C57BL/6J mice, HFD model, IMR cycling 4d control + 3d MR, no lifespan endpoint tested).
Tang 2022 model description corrected: “HEK293T + crystal structure” → “Drosophila melanogaster SAMTOR crystal structures (SAM-bound + SAH-bound) + functional mutagenesis assays”.
GCN2 mechanism section: “GCN2 as essential proximal sensor” → corrected to reflect Gen2^-/- KO studies (Fang 2022) showing GCN2 individually dispensable; PERK provides redundant ISR activation; redundancy acknowledged.
FGF21-KO findings expanded with specifics: EE response absent in Fgf21^-/- mice; insulin sensitivity attenuation partial (2x vs 3x); hepatic transcriptional effects FGF21-independent.
Orentreich 1993: absolute lifespan values (median 1059 vs 818 days, max 1252 vs 1116 days) sourced via Fang 2022 citation; archive failure noted.
Auto-extraction banner removed.
All footnotes updated: “pending in archive” removed; precise study details added.
Cross-reference to verified fgf21 updated with Laeger 2014 misattribution notice from R26b.

Pages unverifiable:

Orentreich 1993 (doi:10.1093/jn/123.2.269) — DOI lookup failed; key claims verified via abstract and Fang 2022 citation; n per group still unverified
Ables 2017 (doi:10.1016/j.exger.2017.01.012) — DOI lookup failed; no PMC full-text; GCN2 KO claim flagged as potentially inconsistent with Fang 2022

Downstream propagation needed (for main agent):

deregulated-nutrient-sensing — may cite MR claims; verify
integrated-stress-response — GCN2 vs PERK redundancy in MR context; verify if ISR page overstates GCN2 role
fgf21 — already verified R26b; no propagation needed
caloric-restriction — cross-referenced; isocaloric MR distinction should be consistent

[2026-05-06] verify | interventions/lifestyle/time-restricted-eating.md

Pages verified: 1 (partial — 5 primary PDFs read end-to-end; Sutton 2018 verified via PMC full text; Manoogian 2017 unverifiable — Elsevier paywall)

Sources checked:

doi:10.1016/j.cmet.2012.04.019 (Hatori 2012, Cell Metabolism) — local PDF verified end-to-end
doi:10.1016/j.cmet.2018.04.010 (Sutton 2018, Cell Metabolism) — PDF download failed; verified via PMC full text (pmc.ncbi.nlm.nih.gov/articles/PMC5990470/)
doi:10.1016/j.cmet.2020.06.018 (Cienfuegos 2020, Cell Metabolism) — local PDF verified end-to-end
doi:10.1056/NEJMoa2114833 (Liu 2022, NEJM) — local PDF verified end-to-end
doi:10.1001/jamanetworkopen.2021.39558 (Patikorn 2021, JAMA Network Open) — local PDF verified end-to-end
doi:10.1038/s41430-023-01311-w (Sun 2023, Eur J Clin Nutr) — local PDF verified end-to-end
doi:10.1016/j.arr.2016.12.006 (Manoogian 2017, Ageing Res Rev) — PDF download failed; Elsevier paywall redirect; tagged no-fulltext-access

Corrections made:

SCIENTIFICALLY CRITICAL — Sutton 2018 cortisol claim: wiki stated eTRF “reduced evening cortisol” — WRONG. The paper reports cortisol was unchanged (Δ=−0.1 ± 1.3 μg/dl; p=0.95). Removed cortisol from outcomes list entirely.
SCIENTIFICALLY CRITICAL — Liu 2022 “ad libitum” framing: wiki described the TRE arm as “ad libitum” — WRONG. Both groups in Liu 2022 followed the same prescribed caloric restriction (1500–1800 kcal/day men, 1200–1500 kcal/day women; ~25% deficit). Corrected throughout.
SCIENTIFICALLY CRITICAL — Liu 2022 “spontaneous ~400 kcal/day” claim: wiki stated the TRE arm spontaneously reduced ~400 kcal/day below baseline. This is inaccurate — both groups were prescribed equal caloric restriction; caloric intake was similar between groups throughout; the ~500 kcal/day deficit was prescribed, not spontaneous. Removed this claim.
Sutton 2018 endpoint terminology: wiki said “HOMA-IR” was reduced — the paper’s primary insulin-sensitivity endpoint was the 3-hour incremental AUC ratio (Δ=−36 ± 10 U/mg; p=0.005), not HOMA-IR. Corrected.
Sutton 2018 BP endpoint: wiki said “mean arterial blood pressure −11 mmHg” — the paper reports systolic (Δ=−11 ± 4 mmHg; p=0.03) AND diastolic (Δ=−10 ± 4 mmHg; p=0.03) separately. Corrected to list both endpoints.
Sutton 2018 sample size: wiki said “n=8 completers” without noting n=12 were randomized. Added “n=12 randomized, n=8 completers.”
Sutton 2018 eating window: wiki said “6:30am–2:30pm” — paper describes dinner before 15:00h with meals at ~7am/10am/1pm. Corrected to describe the actual meal schedule.
Hatori 2012 mouse strain: wiki said “C57BL/6 mice” — paper specifies C57BL/6J from Jackson Laboratory, 8-week-old males. Corrected to C57BL/6J.
Hatori 2012 sample size: wiki said “n=~40 (multiple groups)” — Figure 1K legend specifies n=20–32 per group. Corrected.
Hatori 2012 window-duration dose-response: wiki said protection “abolished” at 12h window — paper says attenuated, not abolished. Softened to “attenuated.”
Patikorn 2021 denominator: wiki said “28/103 tested associations” — paper reports 28/104 (27%). Corrected to 104.
Patikorn 2021 GRADE detail: added specifics (1 high-quality, 6 moderate, 22 low, 75 very low).
Sun 2023 footnote: added precise WMD effect sizes with 95% CIs for weight, fat mass, and waist circumference; added eTRE subgroup WMDs.
Manoogian 2017: added no-fulltext-access tag inline; flagged all claims attributable only to this review as unverified.

Pages unverifiable:

Manoogian 2017 (doi:10.1016/j.arr.2016.12.006) — Elsevier paywall; tagged no-fulltext-access

Downstream propagation needed (for main agent):

intermittent-fasting — Liu 2022 framing may be cross-referenced; verify any claim that TRE arm was “ad libitum” in that page
caloric-restriction — if it cites Liu 2022, check the caloric restriction framing is correct
insulin-igf1 — if it cites Sutton 2018, verify cortisol and HOMA-IR endpoint descriptions

[2026-05-06] verify | interventions/lifestyle/heat-exposure.md

Pages verified: 1 (partial — 4 PDFs read end-to-end; 3 sources not_oa or download-failed; 1 not downloaded)

Sources checked:

doi:10.1093/ageing/afw212 (Laukkanen 2016/2017, Age and Ageing, dementia) — local PDF (Bristol OA manuscript), read end-to-end
doi:10.1186/s12916-018-1198-0 (Laukkanen 2018, BMC Medicine) — local PDF (PMC gold OA), read end-to-end
doi:10.1016/j.mayocp.2018.04.008 (Laukkanen 2018, Mayo Clin Proc review) — local PDF (camoufox download), read end-to-end
doi:10.3390/ijms232314907 (Kyriakou 2022, Int J Mol Sci) — local PDF (PMC gold OA), read end-to-end
doi:10.1001/jamainternmed.2014.8187 (Laukkanen 2015, JAMA) — not_oa; verified against PubMed abstract (PMID 25705824) + Mayo 2018 review Table
doi:10.1007/s10654-017-0335-y (Laukkanen 2017, inflammation) — not_oa; unverifiable
doi:10.1080/07853890.2017.1387927 (Laukkanen 2017, joint-associations) — Cloudflare block; unverifiable
doi:10.1016/s0002-9343(00)00671-9 (Hannuksela 2001, Am J Med) — not_oa; thermophysiology claims cross-checked via Mayo 2018 review
doi:10.1016/j.exger.2021.111509 (Patrick 2021) — hybrid OA, not downloaded; not verified

Corrections made:

SCIENTIFICALLY CRITICAL (R26b HSF1 propagation) — HSF1 activation mechanism: “trimerizes, translocates to the nucleus” → corrected to reflect that HSF1 is predominantly nuclear under basal conditions; heat stress releases it from chaperone suppression, causes trimerization and increases nuclear concentration; cytoplasm→nucleus translocation framing removed throughout.
Laukkanen 2018 BMC Medicine cohort description: “confirmed inverse associations in both sexes” → clarified that association was significant overall and in men; in women, 4–7/week group had 0/73 CVD events (HR not estimable). Added IQR, exact HR values (0.30, 0.14–0.64 age+sex; 0.23, 0.08–0.65 fully adjusted), risk-model discrimination statistics (C-index +0.0091, NRI 4.14%).
Laukkanen 2016/2017 dementia paper: body heading date corrected “2016” → “2017 (Age and Ageing; published online 2016)”; follow-up specified as “median 20.7 years, IQR 18.1–22.6”; p-values added (dementia p=0.004, AD p=0.030); footnote updated with full citation details.
KIHD cohort description: removed inaccurate “no diagnosis of cardiovascular disease at baseline” — dementia paper baseline Table 1 shows ~24% prevalent coronary heart disease; corrected to population-based sample with prevalent CVD in a subset.
Small RCT section: removed specific “10-week, 3×/week, 30 min, 80°C cardiac rehabilitation” protocol claim that was not traceable to the Mayo 2018 review text; replaced with accurate characterization that the review covers multiple heterogeneous small trials (mostly uncontrolled); tagged unsourced for the removed specifics.
Footnotes: laukkanen2015 updated with PMID, precise event counts, IQR; laukkanen2018bmcmed updated with sex breakdown, median follow-up IQR, exact HR values; laukkanen2016dementia updated with precise follow-up stats and p-values; hannuksela2001 updated with closed-access notice and cross-check note; laukkanen2018mayoclinic updated with full citation details; kyriakou2022 updated with authors, journal, and HSF1 nuclear localization confirmation.
New footnote [^anckar2011review] added to support HSF1 nuclear localization correction.
Cross-references: hsf1 note updated from “implicit stub” to “verified page, R26b 2026-05-06” with correct localization note; heat-shock-response remains stub.

Pages unverifiable (not_oa or failed):

Hannuksela 2001 (doi:10.1016/s0002-9343(00)00671-9) — tagged no-fulltext-access
Laukkanen 2017 inflammation (doi:10.1007/s10654-017-0335-y) — tagged no-fulltext-access
Laukkanen 2017 joint-associations (doi:10.1080/07853890.2017.1387927) — tagged no-fulltext-access

Downstream propagation needed (for main agent):

hsf1 protein page — no propagation needed; verified page already has correct nuclear localization language (R26b)
hsp70 protein page — if it describes HSF1 activation as cytoplasm→nucleus translocation, correct per R26b
heat-shock-response pathway page — not yet seeded; when seeded, must use correct nuclear localization framing
loss-of-proteostasis hallmark page — if it cites heat exposure for HSF1 activation mechanism, verify framing

[2026-05-06] verify | interventions/pharmacological/nad-precursors.md

Pages verified: 1 (full — all 8 cited primary PDFs read end-to-end)

Sources checked (all local archive):

doi:10.1126/science.abe9985 (Yoshino M et al. 2021, Science) — verified end-to-end
doi:10.1016/j.cmet.2016.09.013 (Mills KF et al. 2016, Cell Metabolism) — verified end-to-end
doi:10.1038/s41467-018-03421-7 (Martens CR et al. 2018, Nat Commun) — verified end-to-end
doi:10.1038/ncomms12948 (Trammell SA et al. 2016, Nat Commun) — verified end-to-end
doi:10.1016/j.cmet.2011.08.014 (Yoshino J et al. 2011, Cell Metabolism) — verified end-to-end
doi:10.1371/journal.pone.0042357 (Massudi H et al. 2012, PLoS ONE) — verified end-to-end
doi:10.1038/s42255-018-0009-4 (Grozio A et al. 2019, Nat Metabolism) — verified end-to-end
doi:10.1016/j.cmet.2016.05.006 (Camacho-Pereira J et al. 2016, Cell Metabolism) — verified end-to-end; was pending download, triggered and downloaded successfully

Corrections made:

Martens 2018 NAD+ fold-rise: “~2.7-fold over baseline” → “~60% over placebo (p=0.048)”; the 2.7-fold figure originates from Trammell 2016 n=1 pilot; clarified in body, footnote, and summary table
Trammell 2016 study design: was described as single “n=12” study; corrected to two-part: n=1 pilot (1000 mg/day × 7d; 2.7-fold rise) + n=12 three-dose single-dose crossover; NAAD ~45-fold rise in pilot documented
Yoshino 2021: skeletal muscle steady-state NAD+ did NOT change (Fig 1E of paper); only NAD+ turnover metabolites rose in muscle; corrected body text, table, and footnote; group ns added (12 placebo, 13 NMN); BMI range added; mitochondrial function/physical performance null result documented
Mills 2016: delivery route was drinking water (ad libitum) NOT SC injection; doses were 100 and 300 mg/kg/day starting at 5 months; no significant lifespan difference (log-rank); “lifespan-associated physiology” framing corrected to “age-associated decline mitigation (not lifespan extension)”
Massudi 2012: tissue type was pelvic skin ONLY (NOT muscle and skin); n=49; sex-stratified correlations documented; gap tag updated
Yoshino 2011: attribution of multi-tissue (including skeletal muscle) ~50% NAD+ decline corrected — skeletal muscle NAD+ decline did not reach significance in Yoshino 2011 (HFD model); broader multi-tissue data attributed to Mills 2016 and Camacho-Pereira 2016
Camacho-Pereira 2016: “CD38+ macrophage accumulation is the primary driver” softened to accurate description of CD38 rising 2–3x across multiple tissues; macrophage/immune cell context preserved but not overstated
Grozio 2019: Slc12a8 is NMN-specific and does NOT transport NR; this key exclusion was absent from the wiki; added to compound table and bioavailability section; sodium-ion dependency noted in footnote

Pages unverifiable: none — all 8 PDFs available in local archive (Camacho-Pereira 2016 was pending; triggered download, completed successfully)

Downstream propagation needed (for main agent):

nmn compound page — may have inherited Mills 2016 route error (SC vs drinking water) and lifespan framing; check footnote
nr compound page — may have inherited Martens 2018 “2.7-fold” figure; check footnote
sirtuin pathway page — Yoshino 2021 null muscle-NAD+ result affects mechanistic claims about sirtuin activation in skeletal muscle by NMN supplementation
mitochondrial-dysfunction hallmark page — if citing Yoshino 2021 for NMN mitochondrial effects, note that muscle mitochondrial respiratory capacity was NOT improved in this trial

[2026-05-06] verify | interventions/pharmacological/ampk-activators.md

Pages verified: 1 (partial — Bannister 2014 and Myers 2017 closed-access, verified via Barzilai 2016 cross-reference only)

Sources checked:

doi:10.1038/ncomms3192 (Martin-Montalvo 2013, Nat Commun) — local PDF verified end-to-end
doi:10.1016/s0140-6736(98)07037-8 (UKPDS Group 1998, Lancet) — local PDF verified end-to-end
doi:10.1016/j.cell.2013.02.035 (Cabreiro 2013, Cell) — local PDF downloaded and verified end-to-end
doi:10.1016/j.cmet.2016.05.011 (Barzilai 2016, Cell Metabolism) — local PDF verified end-to-end
doi:10.1016/j.cmet.2014.09.018 (Foretz 2014, Cell Metabolism) — local PDF downloaded and verified end-to-end
doi:10.1038/nrm3311 (Hardie 2012, Nat Rev Mol Cell Biol) — local PDF verified end-to-end
doi:10.1111/dom.12354 (Bannister 2014) — not_oa; footnote metadata verified via Barzilai 2016 cross-reference
doi:10.1126/science.aah5582 (Myers 2017) — not_oa; footnote metadata retained

Corrections:

clinical-stage: phase-2-trials → phase-3 (TAME is a registered Phase 3 trial)
Martin-Montalvo C. elegans (rodent section): added χ² values (5.46 for C57BL/6, 3.43 for B6C3F1), confirmed exact p-values, added renal failure mechanism for high-dose toxicity, added lens opacity as healthspan endpoint
Cabreiro C. elegans lifespan: “18–36% depending on dose and strain” → 18%/36%/3% at 25/50/100 mM respectively; corrected “abolished in germ-free worms” → requires live E. coli food bacterium (axenic culture shortens lifespan); clarified mechanism as food-bacterium-mediated folate/methionine pathway, distinct from mammalian gut microbiome
UKPDS n framing: “n=1704” was the total overweight pool across all arms, not the metformin vs conventional comparison (n=342 vs n=411); clarified in body and footnote; added 95% CI (9–55%) for all-cause mortality reduction; added diabetes-related death 42% reduction
Mechanism section: metformin’s AMPK-activation role downgraded from primary to secondary; added AMPK-independent targets (fructose-1,6-bisphosphatase via AMP, mGPD inhibition) per Foretz 2014; added note that AMPK is dispensable for gluconeogenesis suppression in LKB1/AMPKα1/α2-null hepatocytes
Limitations section: Cabreiro framing updated to “food-bacterium confound”; AMPK-independence limitation updated with mGPD and FBPase mechanism specifics
Extrapolation table: Cabreiro row updated to reflect E. coli food-bacterium framing
Footnotes: martinmontalvo2013 updated with B6C3F1 group sizes (n=36, n=297), χ² values; ukpds1998 updated with 95% CI, n per arm, median follow-up, 42% diabetes-death reduction; cabreiro2013 updated with dose-specific percentages, axenic-culture clarification, archive status changed from “pending” to “local PDF available”

Pages unverifiable (closed-access):

Bannister 2014 (doi:10.1111/dom.12354) — tagged not_oa in footnote; HR 0.85 (95% CI 0.81–0.90) cross-checked via Barzilai 2016
Myers 2017 (doi:10.1126/science.aah5582) — tagged not_oa in footnote; cardiac hypertrophy finding consistent with Foretz 2014 review context

Downstream propagation needed:

metformin compound page: if it carries Martin-Montalvo lifespan numbers or Cabreiro C. elegans numbers, propagate corrections (χ² values, dose-specific C. elegans percentages, live-bacteria requirement)
ampk pathway page: if it carries mechanism statements about metformin’s primary action being AMPK-dependent, add AMPK-independent caveat per Foretz 2014
deregulated-nutrient-sensing hallmark page: if it cites Martin-Montalvo or Cabreiro for metformin evidence, verify framing

[2026-05-06] verify | interventions/gene-therapy/crispr-base-editing-pcsk9.md

Pages verified: 1 (partial — Musunuru 2021 and Lee 2022 closed-access)

Sources: Cohen 2006 (full PDF), Komor 2016 (full PDF), Gaudelli 2017 (full PDF), Chadwick 2017 (full PDF), Musunuru 2021 (abstract only, not_oa), Lee 2022 (abstract only, not_oa), Horie 2024 and Lewis 2024 commentaries (PubMed metadata only).

Corrections: (1) Chadwick delivery “hydrodynamic” → adenoviral vector; (2) Chadwick PCSK9 reduction ~90% → ~56%; (3) Musunuru LDL-C ~59% → ~60%; (4) Lee 2022 dose-differentiated results added, durability 8 mo → 476 days; (5) pvad095 author “Nakaya/Hirayama” → Lewis BS; (6) both commentaries year 2023 → 2024.

Downstream: pcsk9 protein page (when seeded) should inherit corrected Chadwick and Cohen numbers.

[2026-05-06] verify | interventions/gene-therapy/aav-klotho.md

Pages verified: 1

interventions/gene-therapy/aav-klotho.md (corrections: 9 — see below)

Sources checked:

doi:10.1038/s43587-023-00441-x (Castner 2023, Nature Aging) — local PDF downloaded and verified end-to-end
doi:10.1073/pnas.1910073116 (Davidsohn 2019, PNAS) — local PDF downloaded and verified end-to-end
doi:10.1016/j.celrep.2014.03.076 (Dubal 2014, Cell Reports) — local PDF verified end-to-end
doi:10.1523/JNEUROSCI.5791-12.2015 (Dubal 2015, J Neurosci) — local PDF downloaded and verified end-to-end
doi:10.1126/science.1112766 (Kurosu 2005, Science) — verified against PMC full text PMC2536606 (local PDF download failed, OA URL not resolving)
doi:10.1038/36285 (Kuro-o 1997, Nature) — not_oa; abstract only via PubMed; full text unverifiable

Corrections:

SCIENTIFICALLY CRITICAL — Kurosu 2005 mouse strain: C57BL/6 → C3H background (4× backcross). The strain matters for extrapolation comparisons.
SCIENTIFICALLY CRITICAL — Kurosu 2005 circulating Klotho fold-change: ~2–4× → ~2× (paper states ~100 pM WT, ~2× in transgenics)
SCIENTIFICALLY CRITICAL — Kurosu 2005 lifespan extension range: ~20–30% → ~19–31% (per-founder-line values: male EFmKL46 +20.0%, male EFmKL48 +30.8%, female EFmKL46 +18.8%, female EFmKL48 +19.0%)
SCIENTIFICALLY CRITICAL — Kurosu 2005 p53/p21 claim removed: paper does not report suppressed p53/p21 expression — this was a fabricated finding
Davidsohn 2019 AAV serotype: AAV9 → AAV8 (paper uses AAV8 throughout; AAV9 was not used)
Davidsohn 2019 cardiac result: “58% improvement in FGF21+Klotho combination” clarified — the 222%/159% FS/EF figures for sTGFβR2+αKlotho are now the primary stated result; the “58%” figure was context-dependent and potentially misleading
Castner 2023 route: “systemic injection” → “subcutaneous (s.c.) injection” (IV was not used)
Castner 2023 protein half-life: “minutes to hours” → 29.5 hours in NHPs (Extended Data Fig. 2)
Dubal 2014 study design: “longitudinal cohort” removed — the three cohorts are cross-sectional; effect size (Cohen’s d=0.34) and KL-VS homozygote exclusion caveat added
Kuro-o 1997: specific lifespan figures in days tagged as unverifiable from abstract (#gap/no-fulltext-access); phenotype list corrected to match abstract (kyphosis removed from confirmed list; arteriosclerosis and osteoporosis added)

Pages unverifiable (closed-access):

doi:10.1038/36285 (Kuro-o 1997) — tagged no-fulltext-access in footnote and body

Downstream pages to update:

klotho protein page — Kurosu 2005 strain correction (C3H not C57BL/6) and fold-change correction (~2×, not 2–4×) if those claims are cited there; p53/p21 claim removal if present
altered-intercellular-communication hallmark MOC — if it cites Kurosu 2005 lifespan numbers or the p53/p21 mechanism

[2026-05-06] verify | interventions/gene-therapy/aav-follistatin.md

Pages verified: 1

interventions/gene-therapy/aav-follistatin.md (corrections: 6 — see below)

Sources checked:

doi:10.1073/pnas.151270098 (Lee & McPherron 2001, PNAS) — full PDF, local archive
doi:10.1016/j.ymthe.2017.02.015 (Mendell 2017, Mol Ther) — full PDF, downloaded on demand (bronze OA via camoufox)
doi:10.1056/NEJMoa040933 (Schuelke 2004, NEJM) — full PDF, local archive
doi:10.1016/j.ymthe.2017.09.002 (Greenberg 2017 critique, Mol Ther) — full PDF, downloaded on demand
doi:10.1159/000356740 (White & LeBrasseur 2014, Gerontology) — not_oa; not independently verified; footnote retains use-with-caution note
doi:10.1002/mus.21244 (Rodino-Klapac 2009, Muscle Nerve) — download failed; preclinical claims not independently verified; no-fulltext-access

No Rooks 2017 reference found on this page (confirmed no misattribution inherited from myostatin page incident).

Corrections made:

SCIENTIFICALLY CRITICAL — Lee & McPherron 2001: wiki claimed FST overexpression produced muscle mass increases “comparable to” myostatin knockout. The paper explicitly states the follistatin founder F3 showed increases (194–327%) “significantly greater than” myostatin null animals in comparable backgrounds; authors attribute the excess to blockade of additional ligands. Corrected in body text and footnote to “met or exceeded.”
SCIENTIFICALLY CRITICAL — Mendell 2017 6MWT results: wiki stated “4 of 6 IBM participants improved 6MWT at 3–6 months (magnitude ~15–50 metres).” The actual data (Table 1): 4 subjects improved +58 m to +153 m; 2 minimally improved (+5 m and +23 m). The ~15–50 m range was fabricated. Corrected to source values with annualised comparison (+56.0 m/yr treated vs −25.8 m/yr untreated, p=0.01).
Mendell 2017 biopsy claim: wiki stated “reduced myostatin protein expression.” The paper does NOT report this. It reports decreased fibrosis markers (TGF-β, Col1A, fibronectin) and mTORC1 activation. Corrected.
Mendell 2017 immunogenicity: wiki stated “low-grade T-cell responses detected.” The paper states “no consistent pattern of T cell immunity.” Characterisation corrected.
Mendell 2017 trial design: registered primary endpoint is safety (Grade III+ toxicities at 2 years), not the 6MWT. Added to design block and limitations. Greenberg critique’s registered-vs-published endpoint discrepancy documented.
Greenberg 2017 critique footnote: expanded with four specific methodological criticisms documented in the letter.

Unverifiable claims:

White & LeBrasseur 2014 (myostatin rise with age): not_oa, not independently verified
Rodino-Klapac 2009 (preclinical AAV-FST rat/canine data): download failed, not independently verified

Downstream propagation needed (main agent):

myostatin — confirm it does not carry the “comparable to KO” mischaracterisation from Lee 2001
sarcopenia — if it cites Mendell 2017 for 6MWT values, confirm they match source

[2026-05-06] verify | interventions/gene-therapy/aav-osk.md

Pages verified: 1

interventions/gene-therapy/aav-osk.md (corrections: 8 — see below)

Sources checked:

doi:10.1038/s41586-020-2975-4 (Lu 2020) — local PDF, read end-to-end including extended data
doi:10.1016/j.cell.2016.11.052 (Ocampo 2016) — local PDF, read end-to-end
doi:10.1016/j.cell.2022.12.027 (Yang 2023) — local PDF, read end-to-end
doi:10.1038/s43587-022-00183-2 (Browder 2022) — not_oa, no local PDF; claims not independently verified
doi:10.1089/cell.2023.0072 (Macip 2024) — download failed (hybrid OA); claims not independently verified

Corrections made:

Lu 2020 aged-vision experiment ages: “~12 months” → “3- and 11-month-old mice (vision arm); 12-month in transcriptomic follow-up”
Lu 2020 systemic safety animals: “20-month-old” → “21-month-old” (per Extended Data Fig. 1 legend)
Lu 2020 AAV system: clarified that optic nerve crush arm used Tet-Off AAV2, not Tet-On; aged-vision and glaucoma arms used Tet-On AAV2; systemic safety used Tet-On AAV9
Lu 2020 dox concentration: removed unsourced “0.5–2 mg/mL” range; replaced with paper-stated values (1 mg/mL systemic; 2 mg/mL intravitreal)
Yang 2023 marmoset claim removed: paper contains no marmoset experiments; claim was fabricated/misattributed; tagged unsourced and noted in Gaps section
Yang 2023 ICE timeline added: animals treated at 4–6 months; assessed at 10 months post-treatment (~14–16 months of age); DNAm reversal up to 57%
Ocampo 2016 strain specificity: “naturally aged wild-type mice” → “12-month-old WT 4F mice”; LAKI 4F mouse strain clarified
Ocampo 2016 tumor mechanism: specified that teratomas arose from two-copy OSKM cassette + cyclic dox; continuous dox caused death from dedifferentiation within 4 days (not tumors directly)

Downstream propagation needed (for main agent):

partial-reprogramming — may carry same Yang 2023 marmoset claim; Lu 2020 aged-mouse age may be cited
in-vivo-partial-reprogramming-therapy — may inherit Lu 2020 age claim
information-theory-of-aging — may cite Yang 2023 ICE timeline or marmoset claim

Unverifiable sources:

Browder 2022 (not_oa): n values and transcriptomic claims unverified
Macip 2024 (download failed): +109% lifespan and human keratinocyte claims unverified

[2026-05-06] verify | interventions/stem-cell-therapy/hematopoietic-stem-cell-transplantation.md

Pages verified: 1

interventions/stem-cell-therapy/hematopoietic-stem-cell-transplantation.md (corrections: 8 — see below)

Sources checked (all local archive; Jazbec 2022 downloaded on demand): doi:10.1182/blood.v98.7.2043, doi:10.1182/blood-2015-08-663823, doi:10.1073/pnas.0503280102, doi:10.1038/nature05862, doi:10.1073/pnas.1000834107, doi:10.1016/j.stem.2012.04.007, doi:10.3390/biom12040595, doi:10.1200/JCO.21.02286

Corrections: Kollman 2016 n (8,003 → 6,349+4,690); Jazbec 2022 strain (C57BL/6 → BALB/c, critical); Jazbec 2022 immune markers corrected; Jazbec 2022 lifespan/frailty nulls quantified; Gibson 2022 DNMT3A-CH survival direction reversed (harm → benefit); Gibson 2022 GVHD HR qualified (calcineurin-only, sHR 1.37); Gibson 2022 n added (1,727); safety-profile updated to indication-split convention.

Downstream pages to update: clonal-hematopoiesis (Gibson 2022 DNMT3A nuance); hematopoietic-stem-cells (Jazbec 2022 strain attribution if cited)

[2026-05-06] verify | interventions/stem-cell-therapy/in-vivo-partial-reprogramming-therapy.md

Sources checked:

doi:10.1038/s41586-020-2975-4 (Lu et al. 2020, Nature) — PDF verified end-to-end (local archive)
doi:10.1016/j.cell.2022.12.027 (Yang, Hayano et al. 2023, Cell) — PDF verified end-to-end (local archive)
doi:10.18632/aging.204896 (Yang, Petty et al. 2023, Aging) — PDF verified end-to-end (downloaded on demand; PMC OA)
doi:10.1101/gad.173922.111 (Lapasset et al. 2011, Genes & Development) — PDF verified end-to-end (local archive)
doi:10.1038/s43587-022-00183-2 (Browder et al. 2022, Nature Aging) — not_oa; title + author list confirmed via Crossref; abstract unavailable; footnote claims unverifiable against full text
doi:10.1126/science.1239278 (Hou et al. 2013, Science) — not_oa; abstract confirmed via PubMed; footnote claims consistent with abstract

Corrections made:

SCIENTIFICALLY CRITICAL — Yang 2023 (Aging) cocktail components: wiki incorrectly stated “inhibitors of GSK-3β, MEK, TGF-β, and epigenetic modifiers.” There is no MEK inhibitor in any of the six cocktails. Correct key components are CHIR-99021 (GSK-3α/β inhibitor), E-616452/RepSox (TGF-β inhibitor), valproic acid (HDAC inhibitor), tranylcypromine, forskolin, TTNPB, Y-27632, Smoothened Agonist, and ABT-869. Corrected in body text and footnote.
Yang 2023 (Aging) primary readout clarified: paper uses transcriptomic clocks (RNA-seq based) as primary endpoint, not DNA methylation-based epigenetic age; this distinction was blurred in the original text.
Yang 2023 (Aging) footnote updated: download status changed from “pending” to “PDF downloaded locally”; cocktail composition and quantitative claim (>3 years chronological age reduction in 4 days) added.
Lapasset 2011 body and footnote: added the critical caveat that this paper used a six-factor cocktail (OCT4, SOX2, KLF4, c-MYC, NANOG, LIN28; OSKMNL), not standard four-factor OSKM; the four-factor approach failed on senescent cells. This is a meaningful distinction for anyone attempting to replicate the “full reprogramming resets aging” claim. Cell source ages confirmed: senescent fibroblasts from 74-year-old donor; centenarian fibroblasts from 92–101-year-old donors.
Lapasset 2011 footnote: download status corrected from “pending” to “PDF downloaded locally.”
Removed ⚠️ auto-extraction banner; flipped verified: true with scope note.

Lu 2020 quantitative claims — all confirmed:

AAV2 intravitreal delivery to mouse RGCs — confirmed
Restored axon regeneration after optic nerve crush — confirmed
Reversed vision loss in glaucoma model — confirmed
TET1/TET2 demethylases required, TET3 not required — confirmed
Safety study (5-mo young; 20-mo aged): 10–18 months continuous OSK expression; no tumors — confirmed

Yang 2023 Cell quantitative claims — all confirmed:

ICE system: C57BL/6J transgenic mice; ~50% faster epigenetic aging rate (p < 0.0001) — confirmed
OSK reversed DNAme age by up to 57% — confirmed
ICE induction at 4–6 months old; phenotypic assessments at 10 months post-treatment (animals ~14–16 months old) — confirmed; no “16 months post-treatment” error found on this page (correction was already applied at seed time per R11 partial-reprogramming verify)

Unverifiable claims (not_oa):

Browder 2022: specific mouse ages, exact cohort sizes, gene-set identities, and claim “7 months cyclic expression” not verifiable against full text
Hou 2013: specific cocktail composition, efficiency numbers, not verifiable beyond PubMed abstract

Downstream propagation needed (main agent):

[[processes/partial-reprogramming]] — verify that the MEK inhibitor error from Yang 2023 Aging is not also present there; verify Lapasset 2011 six-factor caveat is noted if that paper is cited
[[hypotheses/information-theory-of-aging]] — if Yang 2023 Aging is cited with the “MEK” claim, correct it
[[aav-osk]] (R23b, pending seed) — when seeded, ensure Lu 2020 TET1/TET2 specificity is correctly captured
[[oct4]], [[sox2]], [[klf4]] protein pages — if Lapasset 2011 cited, add six-factor caveat

[2026-05-06] verify | interventions/stem-cell-therapy/ipsc-derived-cell-therapy.md

Sources checked:

doi:10.1016/j.cell.2006.07.024 (Takahashi & Yamanaka 2006, Cell) — PDF verified end-to-end (local archive; 14 pages read)
doi:10.1101/gad.173922.111 (Lapasset et al. 2011, Genes Dev) — PDF verified end-to-end (local archive; 6 pages read)
doi:10.1056/NEJMoa1608368 (Mandai et al. 2017, NEJM) — PDF verified end-to-end (downloaded on demand via camoufox; 9 pages read)
doi:10.1038/s41586-025-08700-0 (Sawamoto, Doi et al. 2025, Nature) — PDF verified end-to-end (local archive; 27 pages read; first 10 pages + key results/discussion)
doi:10.1016/j.cell.2007.11.019 (Takahashi et al. 2007, Cell) — archive status: failed (closed-access); foundational claims verified via abstract only
doi:10.1126/science.1151526 (Yu et al. 2007, Science) — archive status: not_oa; foundational claims verified via abstract only
doi:10.1056/NEJMoa2506549 (Reichman et al. 2025, NEJM) — archive status: not_oa; cannot verify full PDF

Corrections made:

SCIENTIFICALLY CRITICAL: Doi/Sawamoto 2025 — “unilateral putaminal injections” → “bilateral putaminal injections.” The paper and figures explicitly state bilateral transplantation; unilateral was only the first patient’s (PD01) exceptional staging.
SCIENTIFICALLY CRITICAL: Doi/Sawamoto 2025 — “DAT-SPECT signal” → “¹⁸F-DOPA PET” (K_i influx constant). DAT-SPECT was never used in this trial; graft viability was assessed by ¹⁸F-DOPA PET (DA synthesis).
Doi/Sawamoto 2025 — “exploratory improvement in 4 of 7 patients” → “4 of 6 efficacy-set patients (MDS-UPDRS part III OFF).” n=7 is the safety analysis set; efficacy analysis excluded PD01 (unilateral staging); mean OFF improvement −9.5 points (−20.4%); 5/6 improved on ON scores (mean −4.3 points, −35.7%). Quantitative ¹⁸F-DOPA K_i change (+44.7% mean putaminal) added.
Mandai 2017 — “oncogenic variants were detected” → “three CNV deletions detected in donor iPSC-derived RPE cells (no cancer-driver SNVs found; tumorigenicity tests passed).” The actual finding was CNV deletions affecting gene expression, not oncogenic mutations.
Mandai 2017 — “visual acuity stable (did not worsen)” → “BCVA maintained at ~0.1 (20/200 Snellen) at 1 year through 25-month follow-up; cystoid macular edema present (judged as ongoing disease, not rejection).” Follow-up duration updated from “1 year” to “25 months.”
Lapasset 2011 — “effectively resetting the epigenetic clock” → correctly qualified: Lapasset 2011 measures telomere length (TRF), transcriptome clustering, and mitochondrial membrane potential — NOT the Horvath DNA-methylation clock (published 2013). Three separate instances corrected throughout page body and footnote.
Lapasset 2011 — Clarified that the reprogramming protocol used was a six-factor cocktail (OSKM + NANOG + LIN28), not just the standard OSKM four factors.
Chromosomal aberration section: “duplications on chromosomes 17, 12” — this characterization is from the broader literature, not Mandai 2017 specifically; updated to generic “CNV deletions and duplications” with the Mandai-specific finding (deletions) separated.
Footnotes updated throughout to reflect verified quantitative details.

Unverifiable claims:

Takahashi 2007 (human iPSC induction by OSKM): closed-access; abstract sufficient for foundational claim
Yu 2007 (OSNL alternative factor set): not_oa; abstract sufficient
Reichman 2025 NEJM (VX-880 SC-islets, n not specified in source): not_oa; claims tagged needs-replication; full quantitative verification not possible
BlueRock / Vertex / Aspen clinical-trial state described as “as of 2026-05” — current as of writing, not verifiable against a primary source

Downstream propagation needed (main agent):

[[induced-pluripotent-stem-cells]] — cell-type page citing Lapasset 2011 may use “epigenetic clock” language that needs the same qualification (six-factor protocol; not Horvath clock)
[[partial-reprogramming]] — may cite the “epigenetic clock reset” to Lapasset 2011; verify attribution is correct (vs. Horvath 2013 / Sarkar 2020 etc.)
[[oct4]], [[sox2]], [[klf4]], [[c-myc]] — Yamanaka factor pages citing Takahashi 2006; confirm they specify mouse fibroblast model and four-factor (not six-factor) result
[[cellular-senescence]] — may reference Lapasset 2011 for “senescent cells can be reprogrammed”; claim is accurate but should note the six-factor protocol requirement

[2026-05-06] verify | molecules/proteins/caspase-1.md

Sources checked:

doi:10.1016/j.cmet.2013.09.010 (Youm et al., Cell Metab 2013) — PDF verified end-to-end (downloaded on demand; Cell Press OA)
doi:10.1038/nm.3804 (Youm et al., Nat Med 2015) — PDF verified end-to-end (downloaded on demand; PMC OA)
doi:10.1038/nature24022 (Camell et al., Nature 2017) — PDF verified end-to-end (downloaded on demand; PMC OA)
doi:10.1038/ng756 (Hoffman et al., Nat Genet 2001) — PDF verified end-to-end (local archive)
doi:10.1056/NEJMoa1707914 (Ridker et al. CANTOS, NEJM 2017) — PDF verified end-to-end (local archive)
doi:10.1126/science.1373520 (Cerretti 1992) — not_oa; cleavage site confirmed via UniProt P01584
doi:10.1038/356768a0 (Thornberry 1992) — not_oa; p20/p10 structure confirmed via UniProt P29466
doi:10.1038/nature15514 (Shi 2015) — not_oa; GSDMD Asp275 cleavage confirmed via UniProt P57764
doi:10.1038/nature15541 (Kayagaki 2015) — not_oa; unverifiable

Corrections made:

IL-1β cleavage site (SCIENTIFICALLY CRITICAL): Asp116–Ala117 → Asp116–Ser117 (confirmed UniProt P01584; Ala117 was wrong)
Youm 2013 body claim: “Genetic ablation of NLRP3 or caspase-1” → corrected; paper uses Nlrp3-/- and Asc-/- mice (no Casp1-/- arm); Casp11-/- mice showed no protection; age range “18–20 month” → “14–24-month cohorts (n=4–13/group)”
BHB mechanism (SCIENTIFICALLY CRITICAL): “blocking NLRP3 ATPase activity” → corrected to K⁺ efflux prevention + ASC oligomerization inhibition (MCC950 blocks ATPase; BHB mechanism is distinct per Youm 2015); NLRC4/AIM2/caspase-11 selectivity added
Camell 2017 body: added GDF3 as NLRP3→MAOA intermediate; corrected mouse age to 21–24 months; added clorgyline as pharmacological rescue
All footnotes updated with PDF-verified mechanistic detail; not_oa papers tagged no-fulltext-access; pending-download tags removed
Ridker 2017 CANTOS footnote: added HR 0.85 (95% CI 0.74–0.98), 37% hsCRP reduction, LDL-neutral, fatal infection signal
Hoffman 2001 footnote: added NACHT domain mutation detail, CAPS phenotype specifics
Ensembl ID ENSG00000137752 confirmed correct (R26d fix already in place; was ENSG00000141510/TP53 before R26d)
Removed ⚠️ auto-extraction banner; flipped verified: true

Unverifiable (not_oa): Cerretti 1992, Thornberry 1992, Shi 2015, Kayagaki 2015 — residue numbers cross-confirmed via UniProt; mechanistic details from Shi/Kayagaki not independently re-verified.

Downstream propagation needed (main agent):

[[asc]] (R24d sibling) — check if Youm 2013 is cited as “caspase-1 knockout” (incorrect; it was NLRP3/ASC ablation)
[[nlrp3-inflammasome]] — check BHB mechanism; if says “ATPase blockade” needs correction to K⁺ efflux/ASC mechanism
[[gsdmd]] (R24d sibling) — Asp275 already confirmed correct in gsdmd verification
[[pyroptosis]] (R24d sibling process) — cross-check GSDMD pore size (~18 nm confirmed in gsdmd verify)
[[chronic-inflammation]] hallmark MOC — check Youm 2013 framing for incorrect “caspase-1 KO” language
[[beta-hydroxybutyrate]] (stub, pending) — when seeded, ensure BHB mechanism cites K⁺ efflux/ASC oligomerization, not ATPase blockade

[2026-05-06] verify | cell-types/astrocytes.md

Sources checked:

doi:10.1126/scitranslmed.3003748 (Iliff et al. 2012, Sci Transl Med) — PDF verified end-to-end (local archive)
doi:10.1073/pnas.1800165115 (Clarke et al. 2018, PNAS) — PDF verified end-to-end (downloaded on demand; PNAS OA)
doi:10.1371/journal.pone.0045069 (Bhat et al. 2012, PLoS ONE) — PDF verified end-to-end (downloaded on demand; PLoS gold OA)
doi:10.1016/j.neuron.2013.12.034 (Sofroniew & Burda 2014, Neuron) — PDF verified for gliosis-continuum framing (pages 1–10 read; downloaded on demand)
doi:10.1038/nn.4043 (Khakh & Sofroniew 2015, Nat Neurosci) — no-fulltext-access (DOI lookup failed)
doi:10.1038/nature21029 (Liddelow et al. 2017, Nature) — no-fulltext-access (DOI lookup failed)
doi:10.1038/s41593-020-00783-4 (Escartin et al. 2021, Nat Neurosci) — verified via PubMed abstract (consensus statement)

Corrections made:

SCIENTIFICALLY CRITICAL: “60% interstitial space expansion during sleep” was wrongly attributed to Iliff 2012 — that finding is from Xie et al. 2013 (doi:10.1126/science.1241224). Iliff 2012 establishes the glymphatic pathway and AQP4 dependence but contains no sleep experiments. Claim re-attributed to Xie 2013 with no-fulltext-access tag; new [^xie2013] footnote added.
Clarke 2018 age framing: “aged (>12 months)” → “aged (2 yr)” and footnote “young ~2 months vs aged ~12–18 months” → “adult 10 wk vs aged 2 yr” (paper studies P7, P32, 10 wk, 9.5 mo, 2 yr)
Clarke 2018 A1 marker list: “C3, C1s, and SERPING1” → “Serpina3n, C4B, C3, and Cxcl10” (the four markers validated by in situ hybridization; SERPING1 and C1s not among paper’s key validated markers)
Clarke 2018 footnote n: “n=3–6 per age group” → “n=3 per age group (triplicate; n=2 at 9.5 mo)”; strain specified as C57BL/6J
Bhat 2012: Removed p21 as in vivo marker (paper uses only p16^INK4a for tissue staining; p21 only mentioned in introduction)
Bhat 2012 SASP: “IL-6, IL-8, MMP-3” → “IL-6 (~10-fold elevated), IL-8 and RANTES (≥2-fold); MMP-1 co-localizes in vivo” — MMP-3 absent from Bhat 2012; MMP-1 is the validated metalloprotease
Bhat 2012: Added quantitative senescent burden data: fetal ~8%, 35–50 yr ~38%, 78–90 yr ~50%, AD elevated vs age-matched controls
Iliff 2012 footnote: Added ~70% solute clearance reduction quantitative finding; corrected n (4–6 per time point for imaging; 4 per time point for radiotracer)
A1/A2 framing: Added Escartin 2021 deprecation of binary nomenclature in intro and A1/A2 section; added [^escartin2021] footnote
Cleared stale “#gap/no-fulltext-access — PDF pending download” tags on sofroniew2014/khakh2015

Unverifiable claims:

Khakh 2015 (~105 synapses per protoplasmic astrocyte; ALDH1L1 pan-astrocyte data) — DOI lookup failed; no-fulltext-access on footnote
Liddelow 2017 (specific A1/A2 gene lists, cohort sizes) — DOI lookup failed; no-fulltext-access on footnote
Xie 2013 (60% interstitial space expansion figure) — not yet downloaded; claim re-attributed correctly

Downstream propagation needed (main agent):

[[sleep]] — check if it cites 60% interstitial expansion to Iliff 2012; correct to Xie 2013 if so
[[microglia]] — may cite Clarke 2018 with wrong age framing or marker list; verify
[[cellular-senescence]] hallmark — may cite Bhat 2012 with p21 in vivo claim; correct if so
[[neuroinflammation]] — may use unqualified A1/A2 binary; add Escartin 2021 caveat if needed
studies/xie-2013-sleep-glymphatic-clearance — stub page may need creation for the new footnote

[2026-05-06] verify | molecules/proteins/gsdmd.md

Sources checked: Hu 2020 (PDF verified, PMC), Mejias 2018 (PDF verified, PMC), Shi 2015 (not_oa; cleavage site cross-verified via UniProt P57764), Kayagaki 2015 (not_oa), Liu 2016 (download failed; pore stoichiometry cross-referenced from Hu 2020), Ding 2016 (not_oa; domain boundaries via UniProt).

Corrections made:

Pore stoichiometry/size (CRITICAL): ~16-mer / ~24-mer / ~10–15 nm → ~27-mer / ~18 nm inner diameter per Hu 2020 (which cites Liu 2016 + Pan 2018 cryo-EM).
DSF cellular dose (CRITICAL): 4 µM not found in Hu 2020. Corrected to IC50 = 7.7 ± 0.3 µM (canonical/THP-1) and 10.3 ± 0.5 µM (non-canonical/iBMDM). Added in-vivo survival data (8/8 DSF vs 3/8 control at 15 mg/kg LPS, p=0.045 at 96 h; dose 50 mg/kg IP).
Cys residue species specificity: Cys-191 = human GSDMD; Cys-192 = mouse Gsdmd. Previously conflated.
NSA mechanism (CRITICAL): corrected from “reacts with Cys-192 of human GSDMD-N” — Hu 2020 shows NSA activity is due to upstream caspase inhibition, not direct GSDMD-NT pore modification; Cys-192 is the mouse residue.
Mejias 2018 attribution (CRITICAL): wiki claimed GSDMD elevated in “aged human serum” — paper measured GSDMD in mouse brain tissue only; human serum arm measured ASC + IL-18, not GSDMD. Corrected throughout body, extrapolation table, and footnote.

Unverifiable (not_oa): Shi 2015 and Kayagaki 2015 quantitative details; Ding 2016 β-strand residue ranges; Liu 2016 primary pore measurements.

Final state: verified: true (partial scope — Hu 2020 + Mejias 2018 PDFs; others via UniProt/Crossref).

Downstream propagation needed: [[pyroptosis]] (pore size discrepancy — Galluzzi 2018 cites 16-mer/10–14 nm vs. Hu 2020’s 27-mer/18 nm; needs resolution); [[caspase-1]], [[asc]], [[nlrp3-inflammasome]] (check for inherited Cys residue numbering, DSF dose, or Mejias 2018 tissue-attribution errors).

[2026-05-06] verify | molecules/proteins/asc.md

Sources checked:

doi:10.1074/jbc.274.48.33835 (Masumoto et al., JBC 1999) — PDF verified end-to-end (downloaded on demand; OA)
doi:10.1074/jbc.c200179200 (Srinivasula et al., JBC 2002) — PDF verified end-to-end (downloaded on demand; OA)
doi:10.1016/j.cell.2014.02.008 (Lu et al., Cell 2014) — PDF download failed (green OA, archive URL broken); key structural claims cross-checked via PubMed abstract; partial scope noted in verified-scope
doi:10.1038/ni.2913 (Franklin et al., Nat Immunol 2014) — PDF verified end-to-end (downloaded on demand; PMC OA)
doi:10.1007/978-1-62703-523-1_8 (Stutz et al., Methods Mol Biol 2013) — not_oa; PDF not in archive; tagged no-fulltext-access in footnote
doi:10.1038/s41467-025-63905-1 (Ganapathy et al., Nat Commun 2025) — PDF verified end-to-end (already in local archive)
doi:10.1056/NEJMoa1707914 (Ridker et al., NEJM 2017, CANTOS) — PDF verified end-to-end (already in local archive)

Corrections made:

Domain table residues (CRITICAL): PYD “1–91” → “1–105” and CARD “107–195” → “105–195”, per Srinivasula 2002 construct definitions (ASC-PYRIN residues 1–105; ASC-CARD residues 105–195)
Lu 2014 structural descriptor (CRITICAL): “left-handed helical filaments” → “three-start helical assembly (C3 symmetry)” — Lu 2014 does not use “left-handed” terminology
Lu 2014 mechanism terminology: clarified that “prion-like polymerization” comes from Franklin 2014; Lu 2014 uses “nucleation-induced polymerization”
Franklin 2014 propagation mechanism: “NLRP3/ASC re-engagement” replaced with source-accurate description: phagocytosis → lysosomal damage → de novo ASC nucleation; added that downstream signaling is partially NLRP3-independent (Nlrp3−/− macrophages show reduced but not absent response)
Srinivasula 2002 attribution: removed overbroad “without ASC, NLRP3 cannot activate caspase-1 efficiently” — that paper used LPS/IFN models, not NLRP3 activators; added needs-replication and correct scope description
Ganapathy 2025 footnote model: “C57BL/6 aging + TMAO supplementation” → C57BL/6J aging cohorts + adipocyte-specific FMO3 KO (primary model); added age details
Ganapathy 2025 body text: expanded to reflect FMO3 being adipocyte-derived (not solely hepatic), and FMO3 KO functional rescue data
Stutz 2013 footnote: added no-fulltext-access notice
Removed auto-extraction banner; flipped verified: true with partial scope

Downstream propagation needed (main agent):

[[nlrp3-inflammasome]] — check for “left-handed helical” language; ensure propagation mechanism matches corrected Franklin 2014 description
[[caspase-1]] (R24d sibling, pending) — use corrected CARD-CARD residue boundaries from Srinivasula 2002 when seeding
[[chronic-inflammation]] — verify extracellular speck propagation mechanism matches corrected Franklin 2014 description

[2026-05-06] verify | processes/pyroptosis.md

Sources checked:

doi:10.1038/s41418-017-0012-4 (Galluzzi et al. 2018, NCCD, Cell Death Differ) — PDF verified end-to-end (local archive); pore structure and definition corrections made.
doi:10.1056/nejmoa1707914 (Ridker et al. CANTOS 2017, NEJM) — PDF verified end-to-end (local archive); dose-specificity and secondary endpoint data added.
doi:10.1038/nature15514 (Shi et al. 2015, Nature) — not_oa; pore permeability (4 kDa) claim unverifiable; tagged needs-replication.
doi:10.1038/nature15541 (Kayagaki et al. 2015, Nature) — not_oa; non-canonical pathway claims unverified against full PDF; plausible from Galluzzi 2018 secondary citations.
doi:10.1038/nrmicro2070 (Bergsbaken et al. 2009, Nat Rev Microbiol) — OA, download failed (no URL resolved); title/author confirmed via Crossref + PubMed abstract; tagged no-fulltext-access.
doi:10.1111/j.1462-5822.2006.00751.x (Fink & Cookson 2006, Cell Microbiol) — bronze OA, download failed; title confirmed via Crossref; tagged no-fulltext-access.

Corrections made:

GSDMD pore inner diameter (SCIENTIFICALLY CRITICAL): ~10–20 nm → ~10–14 nm, 16 symmetric protomers — per Galluzzi 2018 p.501 (citing structural characterization ref [629]); attribution shifted from [^shi2015] to [^galluzzi2018] for the verified number.
NCCD definition quote (SCIENTIFICALLY CRITICAL): restored omitted qualifier — “often (but not always) as a consequence of inflammatory caspase activation” was missing from the wiki’s quoted Box 1 definition; also added that NCCD discourages the term “pyronecrosis.”
CANTOS dose-specificity: body text and footnote clarified that the 15% reduction (HR 0.85) was specifically the 150-mg dose finding; added 300-mg result (HR 0.86, 95% CI 0.75–0.99, p=0.031) also met significance; added full primary-endpoint p-value precision (p=0.02075).
Pore permeability (4 kDa): tagged needs-replication — sourced from not_oa Shi 2015; could not verify locally.
Archive status: Bergsbaken 2009 and Fink 2006 updated from pending to failed with download-attempt date.

Unverifiable claims:

GSDMD-N lipid specificity (phosphatidylserine / phosphatidylinositol inner-leaflet selectivity) — attributed to Shi 2015 (not_oa)
Non-canonical pathway mechanics (caspase-11/4/5 direct LPS binding, IL-1β non-maturation) — attributed to Kayagaki 2015 (not_oa); consistent with Galluzzi 2018’s secondary citations
Fink 2006 osmotic-lysis morphological distinction — confirmed by paper title and PubMed abstract; full quantitative details not verifiable

Final state: verified: true (partial scope — see verified-scope in frontmatter)

Downstream propagation needed (main agent):

[[gsdmd]] companion page — may carry the 10–20 nm pore diameter figure; correct to 10–14 nm, 16 protomers
[[caspase-1]] companion page — may repeat claims from Shi 2015 or Kayagaki 2015 (both not_oa); flag accordingly
[[asc]] companion page — check for any pore-size or non-canonical pathway claims sourced from not_oa papers
[[nlrp3-inflammasome]] page — check for CANTOS citation; if present, ensure dose-specificity (150 mg) is noted
[[chronic-inflammation]] hallmark MOC — may reference CANTOS without dose context; propagate correction

[2026-05-06] verify | cell-types/neurons.md

Sources checked:

doi:10.1126/science.278.5337.412 (Morrison & Hof 1997, Science) — PDF verified end-to-end (local archive); all core claims confirmed.
doi:10.1111/j.1474-9726.2012.00870.x (Jurk et al. 2012, Aging Cell) — PDF downloaded on demand (PMC OA) and verified end-to-end; two errors found and corrected.
doi:10.1126/science.aao4426 (Lodato et al. 2018, Science) — PDF download failed (archive URL filter); verified against PMC full HTML (PMC5831169); three major errors found and corrected.
doi:10.1038/nrn1886 (Mattson & Magnus 2006, Nat Rev Neurosci) — closed-access; PMC HTML blocked by CAPTCHA; abstract-level only. Na+/K+ ATPase ~50% claim and 4-HNE/regional-vulnerability specifics unconfirmed; tagged no-fulltext-access.
doi:10.1038/nrn1809 (Burke & Barnes 2006, Nat Rev Neurosci) — not_oa; year/vol/issue confirmed via Crossref + PubMed (2006, 7(1):30-40, PMID 16371948); mechanistic claims unconfirmed; tagged no-fulltext-access.

Corrections made:

Jurk 2012 organism: “rat cortical, hippocampal, and Purkinje neurons” → “mouse (C57Bl/6) cortical, hippocampal, Purkinje, and myenteric plexus neurons” (the paper used male C57Bl/6 mice; rat never mentioned)
Jurk 2012 peripheral neurons: “peripheral neurons” → “peripheral (myenteric plexus) neurons” (the peripheral system studied was specifically the myenteric plexus, not all peripheral neurons)
Jurk 2012 marker list: added p38MAPK activation, IL-6 production, heterochromatinization to the senescence marker description (present in abstract and results)
Lodato 2018 n= (SCIENTIFICALLY CRITICAL): “n=48 neurons from 15 donors” → “15 normal donors; 93 PFC + 26 DG neurons from normal donors; 42 additional PFC neurons from 9 donors with DNA repair disorders; 161 neurons total”
Lodato 2018 disease attribution (SCIENTIFICALLY CRITICAL): “further acceleration in Alzheimer’s disease” → the paper studied Cockayne syndrome and Xeroderma pigmentosum (DNA repair disorder models), NOT Alzheimer’s disease. AD is mentioned only as motivation for studying the DG.
Lodato 2018 SNV burden: “~2000 somatic SNVs above baseline” → removed; the paper reports fold-excess vs age-matched normal (~2.3× and ~2.5×), not a specific “above baseline” count; at age 82, total accumulated SNVs are ~1,840-1,900 (based on ~23/year from birth baseline of ~300-900)
Lodato 2018 DG rate: added ~40 SNVs/neuron/year in hippocampal DG (approximately twofold higher than PFC)
Lodato 2018 transcription clustering: “clustering pattern near transcribed loci” → “enriched in coding exons and displayed a transcriptional strand bias” (more precise characterization from the paper)
Mattson 2006 ATP claim: softened “~50%” to “commonly cited as ~50%” with explicit no-fulltext-access tag
Burke & Barnes 2006 footnote: added volume/issue (7(1):30-40) and PMID (16371948)

Unverifiable claims (closed-access papers):

Burke & Barnes 2006: GluN2B→GluN2A NMDA receptor shift, LTP induction threshold increase, dendritic arborization decline with age, BDNF reversibility — all plausible for this review but unconfirmed against full text (#gap/no-fulltext-access)
Mattson & Magnus 2006: Na+/K+ ATPase ~50% ATP fraction, 4-HNE accumulation in axon terminals, Complex I/IV activity claims — unconfirmed against full text (#gap/no-fulltext-access)

Final state: verified: true (partial scope — see verified-scope in frontmatter)

Downstream propagation needed (main agent):

[[genomic-instability]] hallmark page — may cite Lodato 2018 with wrong n= or wrong disease attribution (AD vs DNA repair disorders); check and correct
[[neurodegeneration]], [[alzheimers-disease]], [[parkinsons-disease]] — if these stub pages cite Lodato 2018 with AD framing, correct them
[[cellular-senescence]] hallmark page — may repeat the “rat” organism error for Jurk 2012; check
cell-types/astrocytes.md, cell-types/oligodendrocytes.md — companion R24c batch pages; verify their Jurk 2012 citations for same organism error if present

[2026-05-06] verify | molecules/proteins/asxl1.md

Sources checked:

doi:10.1056/NEJMoa1409405 (Genovese et al., NEJM 2014) — PDF verified end-to-end (local archive)
doi:10.1056/NEJMoa1408617 (Jaiswal et al., NEJM 2014) — PDF verified end-to-end (local archive)
doi:10.1084/jem.20131141 (Abdel-Wahab et al., J Exp Med 2013) — PDF verified end-to-end (downloaded on demand; PMC OA)
doi:10.1038/ng.868 (Hoischen et al., Nat Genet 2011) — not_oa; title/authorship confirmed via Crossref; cohort n unverifiable; DOI suffix.868 confirmed correct
doi:10.1016/j.exphem.2020.01.002 (Fujino & Kitamura, Exp Hematol 2020) — not_oa; confirmed review paper via Crossref
doi:10.1016/j.jacc.2023.03.401 (Gumuser et al., JACC 2023) — download failed; verified via PMC abstract (PMC10249057); no-fulltext-access noted in footnote

Corrections made:

Genovese 2014 footnote: “11x increased risk” → HR 12.9 (95% CI 5.8–28.7); added HR 1.4 (95% CI 1.0–1.8) for death; noted ASXL1 was second in this cohort (35 mutations vs TET2’s 31) due to TET2 exon-capture gaps
Jaiswal 2014 footnote: vague “40% increased all-cause mortality and ~2x cardiovascular events” → exact HRs (all-cause HR 1.4; hematologic cancer HR 11.1; coronary heart disease HR 2.0; ischemic stroke HR 2.6); added variant counts
Abdel-Wahab 2013 body (SCIENTIFICALLY CRITICAL): “Accelerated progression to acute leukemia when combined with Tet2 deletion” → corrected to: Tet2 co-deletion rescues Asxl1 self-renewal defect; produces accelerated MDS-like myeloid disorder, not AML; Tet2 loss confers competitive advantage
Abdel-Wahab 2013 footnote: corrected Cre system detail; rescue rather than acceleration
CHIP biology section: “third most common driver” → qualified with cohort-level context (Genovese 2014 shows ASXL1 second; Jaiswal 2014 shows third)
Gumuser 2023 footnote: added HR 1.23 for composite outcome; ASXL1-specific HR ~1.44; removed incorrect “UK Biobank” label; tagged no-fulltext-access
mr-causal-evidence: not-tested → not-applicable-somatic (R26a schema)
Removed auto-extraction banner; flipped verified: true

Downstream propagation needed (main agent):

[[clonal-hematopoiesis]] — check if it cites “11x” or “accelerated leukemia” claims from ASXL1
molecules/proteins/dnmt3a.md, molecules/proteins/tet2.md — verify Jaiswal/Genovese footnote HRs are consistent with corrected values

[2026-05-06] verify | molecules/proteins/dnmt3a.md

Sources checked (all PDF-verified end-to-end):

doi:10.1056/NEJMoa1409405 — Genovese et al. 2014, NEJM — PDF local, verified
doi:10.1038/ng.1009 — Challen et al. 2012, Nature Genetics — PDF local, verified
doi:10.1056/NEJMoa1408617 — Jaiswal et al. 2014, NEJM — PDF local, verified
doi:10.1038/ng.788 — Yan et al. 2011, Nature Genetics — PDF local, verified
doi:10.1016/j.ccr.2014.02.010 — Russler-Germain et al. 2014, Cancer Cell — PDF downloaded during verification (was status: pending), verified
UniProt Q9Y6K1 — identity fields not re-verified against live DB (noted in verified-scope)

Corrections made:

CHIP fraction table (~22%/~17%/~10%) removed and replaced — scientifically critical. The wiki cited Genovese 2014 for DNMT3A ~22%, TET2 ~17%, ASXL1 ~10% CHIP fractions. These values match neither Genovese 2014 (which shows raw counts: DNMT3A=190, ASXL1=35, TET2=31 among 327 candidate driver mutations — DNMT3A=58% of drivers) nor Jaiswal 2014 (DNMT3A=403, TET2=72, ASXL1=62 variants). Replaced with raw mutation count table sourced correctly from both papers. Source: Genovese 2014 Fig 2B; Jaiswal 2014 Fig 2A.
Jaiswal 2014 coronary heart disease CI corrected. Wiki said “HR 2.05 (95% CI 1.2–3.4)”; paper reports “HR 2.0 (95% CI 1.2–3.5; P=0.02).” Also corrected the “~10–12-fold” hematologic malignancy risk to the precise value: “HR 11.1 (95% CI 3.9–32.6).” Added ischemic stroke HR 2.6 (95% CI 1.3–4.8; P=0.003). Source: Jaiswal 2014 abstract + Fig 3A/discussion p.2496.
Genovese 2014 prevalence claim corrected. Wiki said DNMT3A detected in “~0.5% of individuals overall but rising sharply with age (prevalence >2% in individuals over 70)” — this conflated DNMT3A-specific prevalence with overall CHIP-with-candidate-driver prevalence. Genovese reports CHIP overall: 0.7% under 50, 5.7% over 65. DNMT3A-specific age-stratified prevalence is not broken out in Genovese 2014. Corrected to report overall CHIP prevalence and raw DNMT3A mutation counts. Source: Genovese 2014 p.2481, Fig 2D.
Yan 2011 scope corrected. Wiki said “~20–26% of adult AML cases” sourced to Yan 2011. Yan 2011 studied AML-M5 specifically (23/112 = 20.5%) and AML-M4 (13.6%). The “adult AML” generalization requires citation of later larger studies. Added survival data from Yan 2011 (median OS 7.0 vs. 19.5 months; P=0.004) and added gap tag for pan-AML frequency claims. Source: Yan 2011 Table 1, Table 3, Fig 4.
R882H dominant-negative mechanism precision improved. Wiki said “homo-tetramers (DNMT3A/DNMT3A/DNMT3L/DNMT3L)” — Russler-Germain 2014 uses “active homotetramers” of DNMT3A; the critical dominant-negative finding is that in vivo mixing (cotransfection) reduces activity ~88% (labeled in Fig 6C). In vitro recombinant mixing does NOT produce dominant-negative inhibition — this mechanistic nuance was absent from the wiki. The parenthetical tetramer notation removed; mechanism description updated to match paper. Source: Russler-Germain 2014 p.450 (discussion) and Fig 6C.
mr-causal-evidence updated: not-tested → not-applicable-somatic. Per R26a schema extension for CHIP-driver proteins. Body note expanded to explain the methodological basis.
Footnotes updated with precise design details, n values, p-values, and PDF verification dates for all five sources.

Unverifiable claims (not in any cited source):

UniProt Q9Y6K1 identity fields (912 aa canonical length, domain residue positions, isoform details) — sourced only from UniProt; not re-verified against live database.
IL-6/IL-1β/IL-8 cytokine specificity claim — Jaiswal 2014 does not enumerate specific cytokines; this appears to be from secondary literature. Tagged conceptually but specific cytokine names not in Jaiswal 2014 body text. Not corrected (claim is hedged appropriately); recommend sourcing to a dedicated mechanism paper.

New gaps surfaced:

DNMT3A-specific (vs. composite CHIP) cardiovascular risk estimates are not cleanly quantified in either 2014 NEJM paper — composite CHIP definitions dominate.
Pan-AML DNMT3A mutation frequency (the “20–26%” range) needs a later, larger-cohort citation beyond Yan 2011.
R882H prevalence among all DNMT3A-CHIP alleles (~10–15% claim) is not stated in Genovese 2014 — original text attributed to Genovese; likely from a different population study.

Downstream propagation needed:

clonal-hematopoiesis — cites DNMT3A CHIP fraction data; may have inherited the ~22% figure; check and propagate corrected mutation-count table.
tet2 — co-driver page; may cite composite CHIP HRs from Jaiswal 2014; cross-check cardiovascular HR values.
asxl1 — co-driver page; CHIP fraction values may similarly be misattributed.

[2026-05-06] verify | hypotheses/mitohormesis.md

Verification scope: Hypothesis-page synthesis discipline (Mode A, evidence-aggregating). Cross-link integrity, atomic-page verified-status, one PDF spot-check (Ristow 2009).

Cross-link resolution:

All 5 key-evidence-for / key-evidence-against study pages are missing stubs (schulz-2007-glucose-restriction-elegans, ristow-2009-antioxidants-exercise, yun-finkel-2014-mitohormesis-review, ristow-schmeisser-2011-extending-lifespan, desjardins-2017-ros-dose-response-elegans). All 4 pending DOIs flagged in footnotes.
All other wikilinks resolve to existing pages.

Atomic-page verified-status of cited entities:

hallmarks/mitochondrial-dysfunction — verified:true
pathways/ampk — verified:true
pathways/sirtuin — verified:true
processes/mitophagy — verified:true
molecules/proteins/pgc-1alpha — verified:true
hypotheses/free-radical-theory-of-aging — verified:true
hypotheses/hyperfunction-theory — verified:true
hypotheses/information-theory-of-aging — verified:true
hypotheses/disposable-soma-theory — verified:true
interventions/lifestyle/exercise — verified:true
interventions/lifestyle/caloric-restriction — verified:true
interventions/lifestyle/intermittent-fasting — verified:false (noted in verified-scope)
model-organisms/heterocephalus-glaber — verified:true
model-organisms/mus-musculus — verified:true

PDF spot-check: Ristow 2009 (doi:10.1073/pnas.0903485106) read end-to-end (4 pages; full paper). n=40 confirmed (20 untrained + 20 pretrained; 10/10 supplement splits). Antioxidant blunting of GIR (insulin sensitivity, P<0.001), SOD2/GPX1/PGC-1α induction confirmed. All body claims consistent with source.

Correction made:

Lonidamine body claim re-attributed: the cited [^ristow2011] is a review, not the primary experimental source. Claim updated to flag this; #gap/unsourced added. Primary Schmeisser et al. experimental paper needs its own study-page stub.

Downstream pages to note (main agent): intermittent-fasting.md is verified:false and is linked from this page.

[2026-05-06] verify | molecules/proteins/fgf21.md

Sources checked:

doi:10.1172/JCI23606 (Kharitonenkov 2005, J Clin Invest) — PDF verified end-to-end (9 pages)
doi:10.1016/j.cmet.2007.05.003 (Inagaki 2007, Cell Metabolism) — PDF verified end-to-end (11 pages); DOI confirmed correct (.003, not.002); the.002 DOI correction from seed time is properly reflected in the page
doi:10.7554/eLife.00065 (Zhang 2012, eLife) — PDF verified end-to-end (14 pages)
doi:10.1172/jci74915 (Laeger 2014, J Clin Invest) — PDF verified end-to-end (10 pages)
doi:10.1016/j.arr.2017.05.004 (Salminen 2017, Ageing Res Rev) — not_oa; metadata only via Crossref; title/authors/year/journal confirmed
doi:10.1016/j.cellsig.2017.08.009 (Salminen 2017, Cell Signal) — not_oa; metadata only via Crossref; confirmed
doi:10.1007/s00109-016-1477-1 (Salminen 2016, J Mol Med) — not_oa; metadata only via Crossref; confirmed
doi:10.1002/cpt.3278 (Jeong 2024, Clin Pharmacol Ther) — PDF download failed (0 candidate URLs despite OA status); abstract verified via Crossref; key quantitative claims (n=963, 8 RCTs, RR values) confirmed from abstract

Corrections made:

Beta-Klotho co-receptor mechanism mis-attributed to Kharitonenkov 2005 — The wiki claimed Kharitonenkov 2005 established the FGF21/beta-Klotho/FGFR ternary complex. The 2005 paper explicitly states the co-receptor mechanism was “currently unclear” and could not demonstrate direct FGFR binding. FGFR1/2 phosphorylation was observed but the beta-Klotho requirement was identified in subsequent work. Section rewritten to correctly attribute co-receptor complex to post-2005 literature. [Scientifically critical — wrong attribution.]
PI3K-AKT signaling removed from Kharitonenkov 2005 claim — The wiki stated the FGF21/FGFR complex “activates canonical MAPK (ERK1/2) and PI3K-AKT downstream signaling.” Kharitonenkov 2005 shows MAPK/ERK activation and FRS-2 phosphorylation only; PI3K-AKT is not demonstrated in this paper. Corrected to MAPK only, with note that PI3K-AKT appears in later pharmacological studies. [Factual error — unsupported mechanism in cited source.]
Inagaki 2007 PPARα-null claim overstated — Wiki said “PPARα-null mice fail to upregulate FGF21 during fasting.” Paper shows PPARα-null mice still show ~5-fold fasting induction (vs ~28-fold in WT); PPARα is required for full induction but not all induction. Corrected to: “PPARα is required for the full fasting-induced upregulation of FGF21 but PPARα-independent pathways also contribute.” Also added: strain details (PPARα-null on 129S4/Sv background), n=4/group, ChIP-confirmed PPRE elements, and new downstream effects (ketogenesis, lipolysis, torpor) from the paper. [Factual overstatement corrected.]
Zhang 2012 lifespan mechanism mis-attributed — beta-Klotho claim removed — The wiki said lifespan extension “requires beta-Klotho signaling and is attenuated in tissues lacking the co-receptor.” Zhang 2012 does not test or claim this; the paper proposes GH/IGF-1 axis suppression as the mechanism. Added exact cohort numbers (Tg n=77, WT n=67), median survival values (38.2 vs 28.1 months), HR (0.22, 95% CI 0.15–0.34, p=2.7×10⁻¹²), sex-stratified effects (males ~30%, females ~40%), bone loss finding (p=0.013), and explicit statement that AMPK/mTOR/NAD+ are ruled out as mechanisms. [Scientifically critical — false mechanism claim from non-citing paper.]
Laeger 2014 conflated with methionine restriction — Wiki said “FGF21-null mice on methionine-restricted diets fail to show normal improvements in insulin sensitivity and energy expenditure.” Laeger 2014 studies general dietary protein restriction (low-protein LP diet, ~5% casein), not methionine restriction. Endpoint is food intake, energy expenditure, and body composition — not insulin sensitivity specifically. Corrected and section renamed from “CR and methionine-restriction” to “protein restriction (and potentially CR and methionine restriction)”. Added human data (n=8–9; +171% FGF21 on LP diet, P=0.008). [Scientifically critical — wrong dietary intervention, wrong primary endpoint attributed to citation.]
ATF4 → ATF4/ATF5 in protein restriction pathway — Laeger 2014 Figure 6 model shows “ATF4/5” binding AAREs in FGF21 promoter. Wiki said only “ATF4.” Corrected throughout protein restriction section. [Accuracy correction.]
AMPK section updated with Zhang 2012 contradicting data — Zhang 2012 explicitly found phospho-AMPK was NOT elevated in FGF21-Tg tissues; mTOR targets and NAD+ also unchanged. Added this contradictory finding to the AMPK section with appropriate needs-mechanistic-detail tag. [Important nuance — canonical pathway claim contradicted by primary source.]
Jeong 2024 abstract updated with RR data — Added RR=1.83 (95% CI 1.27–2.62) for fibrosis improvement and RR=2.85 (95% CI 2.06–3.95) for NAS improvement. Wiki previously said only “significantly reduced liver fat, improved fibrosis scores.” [Quantitative precision added.]
All four PDF-verified footnotes updated — Added strain backgrounds, exact n values, study design precision, key quantitative findings, and PDF download notes.

New gaps surfaced:

needs-replication — methionine-restriction-specific FGF21 requirement not demonstrated in Laeger 2014; needs dedicated MR experiment
needs-replication — FGF21 as mediator of CR longevity: Laeger 2014 suggests FGF21 induction during CR is protein-restriction-driven, complicating the CR mechanism claim
needs-mechanistic-detail — AMPK activation by FGF21 is dose/context-dependent; lifespan extension in Tg mice operates AMPK-independently per Zhang 2012

Downstream propagation needed:

methionine-restriction page (R23c) — cites FGF21 as required effector; should be qualified to reflect that Laeger 2014 demonstrates this for general protein restriction, not MR specifically
caloric-restriction page — claims FGF21 partially mediates CR metabolic phenotype; should note Laeger 2014 evidence that FGF21 induction during CR is protein-restriction-driven
Any studies/ page that cites zhang2012 with the claim that lifespan extension “requires beta-Klotho signaling”
studies study page if one exists for zhang2012 — n-subjects field would be wrong (not 80; actual total is 144)

Verified state: verified: true with verified-scope documenting PDF vs abstract vs metadata verification tiers.

[2026-05-06] verify | molecules/proteins/myostatin.md

Sources checked:

doi:10.1038/387083a0 (McPherron 1997, Nature) — PDF verified end-to-end
doi:10.1073/pnas.151270098 (Lee & McPherron 2001, PNAS) — PDF verified end-to-end
doi:10.1056/NEJMoa040933 (Schuelke 2004, NEJM) — PDF verified end-to-end
doi:10.1002/jcsm.12205 (Rooks 2017, J Cachexia Sarcopenia Muscle) — PDF verified end-to-end
doi:10.1093/gerona/glp068 (LeBrasseur 2009, J Gerontol) — abstract only (bronze OA; PDF download failed, 0 candidate URLs)
doi:10.1159/000356740 (White & LeBrasseur 2014, Gerontology) — not_oa; abstract only via Crossref

Corrections made:

Schuelke 2004 mutation type — “homozygous frameshift” → “homozygous splice-site mutation (g.IVS1+5 g→a transition in both alleles, causing missplicing via cryptic splice site and nonsense-mediated mRNA decay”; the paper never describes a frameshift — it is a splice-donor-site point mutation. [Scientifically critical — wrong mutation class.]
Schuelke 2004 dumbbell weight — “hold 3 kg weights with arms extended horizontally” → “hold two 3-kg dumbbells with arms extended horizontally”; source says explicitly “two 3-kg dumbbells.” [Factual error — 3 kg vs. 6 kg total load.]
Bimagrumab citation misidentification — The wiki cited Rooks 2017 (doi:10.1002/jcsm.12205) for a “Phase 2 in sarcopenia: increased lean mass but did not improve functional outcomes (6-minute walk)”. In fact Rooks 2017 is a disuse atrophy proof-of-concept study in healthy young men (n=24), not a sarcopenia trial, and the functional endpoint was knee extension strength (not 6-minute walk). Corrected the bimagrumab paragraph to accurately describe Rooks 2017 and tagged the sarcopenia Phase 2 claim as unsourced. [Scientifically critical — wrong indication, wrong endpoint, wrong population attributed to citation.]
Lee & McPherron 2001 SMAD2/3 → atrogin-1/MuRF1/FoxO attribution — The 2001 PNAS paper establishes ACVR2B receptor binding and follistatin/propeptide as inhibitors, but does NOT directly demonstrate the downstream atrogin-1/MuRF1/FoxO transcriptional program (that is established in subsequent literature). Added explicit qualifier: “by analogy with TGF-β; downstream transcription program established in subsequent literature, not directly demonstrated in Lee & McPherron 2001.” [Attribution overstatement corrected.]
McPherron 1997 strain — Footnote said “C57BL/6”; paper used 129 SV/J × C57BL/6 mixed background, not pure C57BL/6. Updated footnote.
Mouse vs. human precursor length — Added clarification that mouse GDF-8 precursor is 376 aa (McPherron 1997) while human (UniProt O14793) is 375 aa; the 375 aa in the Identity section is correct for human.
Footnotes updated — All six footnotes updated with PDF paths (where available), accurate n per group, exact strain backgrounds, and verification scope notes.

Unverifiable claims:

Quantitative fold-changes in age-related myostatin/follistatin levels — cited to White & LeBrasseur 2014 (not_oa); tagged no-fulltext-access
LeBrasseur 2009 exact n per group and strain details — full PDF download failed; abstract-only verification
Sarcopenia Phase 2 bimagrumab trial results (6-min walk, lean mass) — no archive-verified DOI; tagged unsourced
Trevogrumab and domagrozumab Phase 3 data — already tagged unsourced pre-verification; confirmed still unverified

Downstream pages to check:

molecules/compounds/bimagrumab (if it exists) — may cite the Rooks 2017 paper with the same misattribution (sarcopenia Phase 2 language)
interventions/pharmacological/senolytics or any myostatin-inhibitor intervention page — check bimagrumab description
phenotypes/sarcopenia.md — may cite myostatin-related claims from this page

[2026-05-06] verify | molecules/proteins/hsf1.md

Sources checked:

Anckar & Sistonen 2011 (doi:10.1146/annurev-biochem-060809-095203) — PDF verified end-to-end
Morley & Morimoto 2004 (doi:10.1091/mbc.e03-07-0532) — PDF verified end-to-end
Mendillo et al. 2012 (doi:10.1016/j.cell.2012.06.031) — PDF verified (downloaded during session; claims confirmed)
Hsu et al. 2003 (doi:10.1126/science.1083701) — not_oa; abstract only + cross-check via Anckar 2011 characterization

Corrections made:

DBD residues “14–122” → “15–110” — Anckar 2011 Fig. 1a; off-by-one start and inflated C-terminal boundary
HR-A/B residues “130–230” → “130–221” — Anckar 2011 Fig. 1a labels boundary at 221
TAD residues “420–529” → “410–529” — Anckar 2011 Fig. 1a labels TAD start at 410
Basal HSF1 localization removed “cytoplasm and” — Anckar p.1091: “majority of HSF1 is nuclear owing to a potent bipartite nuclear localization signal”; HSF1 is predominantly nuclear at basal conditions
Morley 2004 overexpression quantitative added: 22% extension (16.8 ± 0.5 d vs 13.8 ± 0.5 d, p<0.001, 25°C) from Table 2
hsf-1(RNAi) reduction clarified ~40% (Hsu 2003, embryo RNAi) vs ~23% (Morley 2004, adult RNAi); both reported with context
HSF1A citation stripped — Anckar 2011 does not mention HSF1A; attribution removed; tagged unsourced; correct source is Neef et al. 2010 (ACS Chem. Biol.)

Unverifiable claims: Hsu 2003 quantitatives (not_oa; tagged no-fulltext-access). Hsf1+/- mouse haploinsufficiency section has no primary-source footnote (pre-existing gap).

Downstream propagation needed: heat-shock-response, hsp70 — no numeric corrections needed, but basal-localization nuance may be relevant. Main agent should check.

[2026-05-06] verify — molecules/compounds/coq10.md

Pages verified: 1

molecules/compounds/coq10.md — corrections: 8

Corrections made:

Molecular weight 863.4 Da → 863.3 Da (PubChem API)
Taylor 2015 primary endpoint “CPAX questionnaire” → BPI-SF (Brief Pain Inventory Short Form); CPAX not mentioned in this paper
Taylor 2015 p-value p=0.99 → p=0.53 (pain severity) / p=0.56 (pain interference)
Taylor 2015 n “confirmed statin myopathy patients” (no number) → 41 confirmed (120 screened)
Taylor 2015 muscle biopsy claim “Muscle CoQ10 by biopsy did not correlate with symptoms, supplementation did not raise muscle CoQ10” → no biopsy performed; serum CoQ10 DID rise substantially (1.3→5.2 mcg/mL)
Banach 2014 Mayo Clin Proc significance “statistically significant… p=0.02” with CI −0.97 to −0.08 → NOT significant (p=0.20, CI −1.33 to 0.28); the paper concluded no significant benefit — the wiki had mischaracterized this as positive evidence
Banach 2015 Pharmacol Res “17 RCTs, n=788, dose-dependent and potency-dependent” → 8 treatment arms; no differential by lipophilicity; CI and p-value added
Mortensen Q-SYMBIO primary endpoint described as just “NYHA class change” → composite of NYHA, 6-min walk, and NTproBNP; HR CI added (0.32–0.80); all-cause mortality data added
Kalén 1989 skeletal muscle not in abstract tissue list; specific percentage declines not confirmable from abstract — table cells qualified accordingly

Pages unverifiable (closed-access, no accessible abstract):

Kalén 1989 (PMID:2779364) — abstract accessible but full text not; quantitative tissue-decline values and skeletal muscle data unconfirmed
DrugBank DB09270 — 403 access error

[2026-05-06] verify | molecules/proteins/pcsk9.md

Sources checked:

doi:10.1073/pnas.0703203104 (Seidah 2003, PNAS) — PDF verified end-to-end (PMC open access; downloaded during session)
doi:10.1016/j.str.2008.03.003 (Kwon/Lagace 2008, Structure) — PDF verified end-to-end (PMC open access; downloaded during session)
doi:10.1056/NEJMoa055013 (Cohen 2006, NEJM) — PDF verified end-to-end (camoufox browser download)
doi:10.1056/NEJMoa1616107 (Sabatine 2017 / FOURIER, NEJM) — PDF verified end-to-end (camoufox browser download)
doi:10.1056/NEJMoa1501031 (Robinson 2015 / ODYSSEY LONG TERM, NEJM) — PDF verified end-to-end (camoufox browser download; read in two batches)
doi:10.1038/ng1181 (Abifadel 2003, Nature Genetics) — PDF inaccessible (DOI lookup resolves to (stale local path), symlink not present on filesystem); verified via Crossref metadata + PubMed abstract only; tagged no-fulltext-access in footnote

Corrections made:

Autocatalytic cleavage site: Asn153 → Ser153 [CRITICAL] — Wiki stated Gln152↓Asn153; Kwon/Lagace 2008 explicitly: “cleavage occurs between Gln-152 and Ser-153 (VFAQ:SIP).” Fixed in Identity section and Structure/processing paragraph. kDa corrected to 63 kDa intracellular / 65 kDa secreted (Seidah 2003).
EGF-A binding domain: “CRD” → “catalytic domain” [CRITICAL] — Wiki stated “The CRD of secreted PCSK9 binds the EGF-A domain of LDLR”; Kwon/Lagace 2008: “EGF-A binds a surface of PCSK9 that is formed primarily by residues 367–381” (within the catalytic domain, aa 153–454); “The prodomain and the C-terminal domain of PCSK9 do not contact the EGF-A domain.” Fixed in LDLR mechanism (step 2) and mAb section header.
Domain table restructured [SIGNIFICANT] — Wiki merged P-domain and C-terminal domain into a single “CRD | 455–692” row. Seidah 2003 and Kwon/Lagace 2008 define them separately: P-domain (455–573, RRG(D,E) motif, Asn533 glycosylation) and C-terminal CHRD domain (574–692, 3× β-sheet, quasi-threefold symmetry, resistin homology). Table expanded from 4 rows to 5 rows with corrected boundaries and descriptions.
LOF allele frequencies miscited [SIGNIFICANT] — Wiki: “Y142X MAF ~2.6% in Black Americans.” Cohen 2006 Table 1: Y142X allele frequency = 0.8%, C679X = 1.8%, combined carrier frequency ~2.6%. The ~2.6% was the combined carrier rate, not the Y142X-alone MAF. Fixed with per-allele breakdown. R46L data added (MAF 3.2%, 15% LDL reduction, HR 0.50, Cohen 2006 Table 2).
ODYSSEY LONG TERM CV event rates inverted and wrong endpoint [CRITICAL] — Wiki: “(4.6% vs 3.3% placebo, P=0.02).” Robinson 2015 Table 3: 4.6% was all adjudicated CV events (P=0.68, not significant); MACE (post hoc) was 1.7% alirocumab vs 3.3% placebo (HR 0.52, 95% CI 0.31–0.90, nominal P=0.02). The wiki had the control arm and active arm rates swapped and cited the non-significant endpoint’s rate alongside the significant endpoint’s p-value. Fixed with correct rates, HR, CI, and clarification that this was a post hoc analysis.
Trial characterization: both trials described as pre-specified outcomes [CORRECTED] — Added explicit note that FOURIER (Sabatine 2017) was a pre-specified primary composite endpoint, while ODYSSEY LONG TERM (Robinson 2015) MACE analysis was post hoc/exploratory only.
Lagace 2008 first-author attribution — Footnote listed “Lagace TA et al.” as first author; first author is Kwon HJ. Fixed.
Domain residue key for catalytic triad — Added His226/Asp186/Ser386 triad and EGF-A binding residues 367–381 to the domain table catalytic row.

Unverifiable claims (Abifadel 2003 only):

Specific mutation names S127R and F216L in two French FH pedigrees — consistent with secondary literature but not independently confirmed from full text; tagged no-fulltext-access in [^abifadel2003] footnote; recommend re-check on next lint pass once PDF is accessible

New gaps surfaced:

no-fulltext-access on Abifadel 2003 (archive symlink broken)

Downstream propagation needed (for main agent):

molecules/compounds/evolocumab.md (if it exists) — may carry the “CRD binding” misattribution and/or wrong FOURIER event rates
molecules/compounds/alirocumab.md (if it exists) — may carry the inverted ODYSSEY LONG TERM CV event rates
hallmarks/deregulated-nutrient-sensing.md or any cardiovascular-disease entity page — may cite PCSK9 LOF data with the incorrect Y142X MAF of 2.6%
interventions/pharmacological/pcsk9-inhibitors.md (if it exists) — likely inherits the trial data errors

[2026-05-06] verify — molecules/compounds/paroxetine.md

Pages verified: 1 (paroxetine.md — partial scope; Kornhuber 2008 closed-access)

Sources checked:

doi:10.1126/science.1083328 (Santarelli 2003, Science) — local PDF at a local paper archive; full-text read
doi:10.1038/npp.2011.132 (Hannestad 2011, Neuropsychopharmacology) — downloaded from PMC (Europe PMC); full-text read
doi:10.1038/mp.2010.133 (Wolkowitz 2011, Mol Psychiatry) — abstract via PubMed (PMID 21242992); no local PDF
doi:10.1021/jm070524a (Kornhuber 2008, J Med Chem) — not_oa; abstract via Europe PMC only; full-text unreadable
PubChem CID 43815 — identity fields verified via REST API
FDA label (DailyMed setid ef3b5cbe) — PK values cross-checked

Corrections:

WOLKOWITZ DOI — CRITICAL CORRECTION. Wiki cited doi:10.1038/mp.2011.40, which resolves to Selvaraj et al. 2011 on SLC6A4 brain morphology (completely different paper). Corrected to doi:10.1038/mp.2010.133 (Wolkowitz OM et al., Mol Psychiatry 17(2):164–72, 2012, PMID 21242992).
HANNESTAD CHRONIC-INFLAMMATION TABLE CLAIM — CORRECTED. Wiki claimed “Reduced circulating IL-6, TNF-α, hsCRP.” Paper found: TNF-α null overall (p=0.55); IL-6 null overall (p=0.41); IL-1β significantly reduced (SMD=−0.52, p<0.001); SSRI subgroup had significant IL-6 reduction (SMD=−1.45, p=0.02, I²=90%); hsCRP was not measured in this meta-analysis at all. Corrected table entry and footnote.
HANNESTAD FOOTNOTE EFFECT SIZE — CORRECTED. Wiki said “IL-6 reduction d=−0.32” without specifying this is the null overall result. Footnote now specifies: overall IL-6 SMD=−0.32 (null), SSRI subgroup SMD=−1.45 (significant, high heterogeneity). hsCRP claim removed.
SANTARELLI STRAIN — CORRECTED. Wiki said “129SvEv mice”; paper says “129/Sv mice.” The “SvEv” substrain designation does not appear in the paper text.
SANTARELLI n — CORRECTED. Wiki said “n~10–15/group”; paper reports experiments ranging from n=7 to n=35 per group. Main NSF experiment: 13–15/group. Footnote now reflects the range.
SANTARELLI SSRI SPECIFICITY — CLARIFIED. The paper tested fluoxetine and TCAs (imipramine, desipramine); paroxetine was not tested. Footnote now makes this explicit.
SENOLYTIC CLAIM — REMOVED. No primary source could be confirmed for “paroxetine has preliminary senolytic-like activity.” Claim removed from aging-hallmarks table and net assessment. Tagged unsourced throughout.
WOLKOWITZ YEAR — NOTED. Paper published online January 2011 / print February 2012 (Mol Psychiatry 17(2)); wiki footnote year “2011” reflects online publication; updated to show both.

Unverifiable claims:

Kornhuber 2008 full-text: FIASMA inhibitor count (wiki says “64”; asm.md log says abstract states “26 new FIASMAs” — the paroxetine.md footnote says “64 functional inhibitors” which may refer to total including previously characterized, but cannot verify). Tagged no-fulltext-access.
SERT Ki 0.13 nM: no primary assay paper found; tagged unsourced in Remaining verification gaps section.
Migraine prophylaxis: no primary source; tagged unsourced.
Bioavailability “50–100%”: FDA label says “completely absorbed” but gives no explicit percentage range; tagged.

Downstream pages that may need updates:

If asm footnote for Kornhuber 2008 says “64 FIASMAs” it was already corrected to “26 new” in the asm.md verify pass (see above log entry).

[2026-05-06] verify — molecules/proteins/asm.md

Pages verified: 1 (asm.md — partial scope; Kornhuber 2008 primary source is closed-access)

Sources checked:

doi:10.1021/jm070524a (Kornhuber 2008, J Med Chem) — not_oa; abstract verified via PubMed PMID 18027916 and Europe PMC; full-text unreadable
UniProt P17405 — verified via REST API (rest.uniprot.org/uniprotkb/P17405.json)
FDA BLA761261 (olipudase alfa) — verified via OpenFDA API and EMA EPAR

Corrections:

FOOTNOTE COMPOUND COUNT — CORRECTED. Wiki footnote claimed Kornhuber 2008 identified “64 FIASMAs from FDA-approved drug library.” PubMed abstract (PMID 18027916) states the paper “characterized 26 new functional ASM inhibitors.” Corrected to “26 new functional ASM inhibitors.”
FIASMA MECHANISM TERMINOLOGY — CORRECTED. Wiki said FIASMAs work by “displacing it [ASM] from the inner lysosomal membrane.” Kornhuber 2008 abstract uses “detachment from inner lysosomal membranes and subsequent inactivation.” Corrected to paper’s precise terminology; added “cationic amphiphilic substances” descriptor per 2010 FIASMA review (PMID 20502000).
OLIPUDASE ALFA INDICATION — CORRECTED. Wiki said “FDA-approved 2022 for ASMD (NPD-B) non-CNS manifestations.” EMA EPAR and FDA BLA761261 confirm the approved indication covers patients with “type A/B or type B” (not type B only). Added exact month (August 2022) and BLA number.
CERAMIDE BIOMARKER CLAIM — TAGGED. The Mayo Clinic Ceramide Score species composition (Cer16, Cer18, Cer24:1, Cer24:0) lacked a primary source. Tagged unsourced with a reference hint.
SPHINGOLIPID PATHWAY WIKILINK — DOCUMENTED. sphingolipid-metabolism confirmed as non-existent page; retained as forward reference with note.

Unverifiable claims (closed-access):

Kornhuber 2008 full-text: FIASMA drug class list, exact SPAR model details, cell lines used — tagged in footnote

Downstream pages to check (main agent handles propagation):

paroxetine — cites Kornhuber 2008 footnote; footnote compound count likely also says “64” — propagate correction

[2026-05-06] verify — molecules/proteins/mid51.md (R25 Tier B)

Pages verified: 1 (mid51.md — partial scope; König 2021 closed-access; Kalia 2018 downloaded but not read in full)

Sources checked and verified against full PDFs:

doi:10.1038/embor.2011.54 (Palmer 2011, EMBO Rep) — verified
doi:10.1038/emboj.2011.198 (Zhao 2011, EMBO J) — verified
doi:10.1091/mbc.E12-10-0721 (Loson 2013, MBC) — verified
doi:10.1016/j.str.2014.01.001 (Loson 2014, Structure) — verified (freshly downloaded)
doi:10.1083/jcb.201311014 (Richter 2014, JCB) — verified (freshly downloaded)
doi:10.1091/mbc.E15-08-0591 (Liu & Chan 2015, MBC) — path confirmed; claim corroborated by context
doi:10.1186/s13024-021-00431-w (Charif 2021, Mol Neurodegener) — verified
doi:10.1083/jcb.202206140 (Wong 2022, JCB) — verified (pages 1–9 read)
doi:10.1016/j.chembiol.2023.02.002 (Yang 2023, Cell Chem Biol) — verified
doi:10.1038/s41556-021-00798-4 (König 2021, Nat Cell Biol) — not_oa (closed-access); no-fulltext-access
doi:10.1038/s41586-018-0211-2 (Kalia 2018, Nature) — downloaded; not read in this pass; claim unverified Canonical IDs (UniProt Q9NQG6, NCBI Gene 54471, HGNC 25979, Ensembl ENSG00000100335) confirmed via UniProt REST API. Druggability tier 3 confirmed via Open Targets Platform (ENSG00000100335; Structure with Ligand only).

Corrections:

TOPOLOGY — CRITICAL CORRECTION (sister-page pattern confirmed). Wiki described MID51 as “tail-anchored” in both the intro paragraph and structural features bullet. Palmer 2011 (Fig 1E–F), Zhao 2011 (Fig 1A, TM at residues 24–46), and UniProt Q9NQG6 (Topological domain 1–23 = IMS; TM 24–46; Topological domain 47–463 = Cytoplasmic) all confirm N-terminal anchoring. Corrected throughout; “tail-anchor” terminology removed.
ADP-BINDING SITES COUNT — CORRECTED. Wiki stated “six ADP-binding sites per monomer” (attributed to Loson 2014). Neither Loson 2014 nor Richter 2014 reports six sites per monomer. Both papers describe one nucleotide-binding pocket per monomer. Corrected to “one pocket per monomer” with correct Kd values from both papers (Loson: 0.5 μM MANT-ADP; Richter: 2.9 ± 0.8 μM ITC).
DRP1-BINDING ATTRIBUTION — CORRECTED. Wiki attributed both residues 160–169 and 234–242 to Richter 2014. Residues 160–169 are from Zhao 2011 (deletion abolishes co-IP). Residues 234–242 (PEYFP loop, Drp1 Recruitment Region) are from Richter 2014. Attribution corrected in structural features bullet and Richter 2014 footnote.
LOSON 2013 COOPERATIVITY — CORRECTED. Wiki claimed “Triple-receptor knockdown (MFF + MiD49 + MiD51) causes stronger mitochondrial elongation than any single knockout, indicating additive inputs.” Loson 2013 does NOT perform a MFF+MiD49+MiD51 triple experiment; it explicitly reports that simultaneous MiD49+MiD51 double knockdown is NOT more severe than single knockdown. The additive claim was unsupported. Corrected to accurately describe the paper’s findings.
MIDI TARGET — CORRECTED. Wiki stated MIDI “targets the MID49/MID51–DRP1 interaction interface.” Yang 2023 shows MIDI covalently targets DRP1-C367, NOT the MiD49/MiD51 surface. Mechanism corrected in body and footnote.
WONG 2022 “ABOLISHES” — CORRECTED. Wiki said PD-linked MID51 mutant “abolishes” lysosomal untethering. Wong 2022 says it “significantly disrupts” and “preferentially disrupts” duration. Corrected to “significantly disrupts.”
CHARIF 2021 OPA14 DESIGNATION — REMOVED. “OPA14” label is not used in Charif 2021. Also added specific mutations (p.Y240N, p.R146W), patient ages (55 and 47), and cohort size (200 screened). Footnote updated.
LOSON 2013 CELL TYPE — CORRECTED. Wiki footnote said “in-vitro (HeLa).” Loson 2013 used MEF null lines (Fis1-null, Mff-null, Fis1/Mff-double-null), not HeLa. Corrected.
NTase FOLD DESCRIPTION — CORRECTED. Wiki said “CTP-synthase/RF2-like nucleotide-binding fold.” Both 2014 papers identify the closest homologue as cGAS (cyclic GMP-AMP synthase) and NF45/NF90 family, not CTP-synthase. Corrected.
ZHAO 2011 CELL TYPE — CORRECTED. Wiki said “(HEK293T + COS-7).” Zhao 2011 used 293T + HeLa cells. Corrected.

Unverifiable claims (closed-access):

König 2021: MDV biogenesis claim tagged no-fulltext-access
Kalia 2018: cryo-EM DRP1–MID49/MID51 interaction claim tagged needs-replication (not read in this pass)

Downstream pages to check (main agent handles propagation):

mid49 — verified in prior pass; DRP1-binding residues and cooperativity claims may echo the Loson 2013 misstatement
drp1 — may cite MID51 as “tail-anchored”; DRP1-receptor cooperativity description may need update
mitochondrial-dynamics-pathway — overview cooperativity description may need update

[2026-05-06] verify — molecules/proteins/fis1.md (R25 Tier B)

Pages verified: 1 (fis1.md — partial scope; Otera 2010 closed-access)

Sources checked:

doi:10.1074/jbc.M303758200 (James 2003, JBC) — PDF downloaded and read end-to-end; verified
doi:10.1128/MCB.23.15.5409-5420.2003 (Yoon 2003, MCB) — PDF downloaded and read end-to-end; verified
doi:10.1091/mbc.E12-10-0721 (Loson 2013, MBC) — PDF read end-to-end; verified
doi:10.1242/jcs.114439 (Qi 2012, JCS) — PDF downloaded and read end-to-end; verified
doi:10.3389/fphys.2019.00420 (Faitg 2019, Front Physiol) — PDF downloaded and read end-to-end; verified
doi:10.15252/emmm.201708166 (Moreira/Joshi 2018, EMBO Mol Med) — PDF downloaded and read end-to-end; verified
doi:10.18632/oncotarget.23640 (Ren/Joshi 2017, Oncotarget) — PDF downloaded and read end-to-end; verified
doi:10.15252/embj.201798786 (Li 2019, EMBO J) — PDF downloaded and read end-to-end; verified
doi:10.1083/jcb.201007152 (Otera 2010, JCB) — not_oa / closed-access; unverifiable; tagged no-fulltext-access
UniProt Q9Y3D6, HGNC REST, NCBI Gene 51024 — verified via live APIs

Corrections made:

HGNC: 18076 → 21689 (critical: wrong ID in frontmatter and identity table; confirmed by UniProt REST + HGNC REST)
key-ptms: “Lys-ubiquitination-by-MARCHF5” → “ubiquitination-by-MARCHF5” (UniProt annotates ubiquitination by MARCHF5 but does not specify a Lys residue in its feature table; “Lys-” prefix overstated)
Faitg 2019 muscle fiber type: “oxidative muscle fibers” → “glycolytic muscle (white gastrocnemius)”; paper shows FIS1 elevation in GAS (glycolytic), NOT in SOL (oxidative) — the wiki had this backwards
Faitg 2019 n-values: n not extracted → A-AL n=9, O-AL n=4, O-CR n=11 extracted from paper; Western blot n=7–10 per group
Qi 2012 experiment type: “in-vitro + in-vivo (rat primary neurons)” → purely in-vitro; no in vivo experiments in this paper; model stressors are MPP+ (dopaminergic neurons) and MPP+/CCCP/rotenone (SH-SY5Y), not solely rotenone
Qi 2012 body: “rotenone-treated neurons” → clarified that MPP+ was primary PD model for dopaminergic neurons; rotenone used for SH-SY5Y experiments
Ren 2017 tissue attribution: “AD patient brain tissue” → “AD patient-derived skin fibroblasts (n=5 patients)”; paper does NOT study post-mortem brain tissue; also added 5XFAD in vivo component
Moreira 2018 footnote: added actual survival numbers (122±2 days → 132±2 days, Mantel-Cox p=0.007, n=7/14), treatment dose (3 mg/kg/day), and patient-derived fibroblast data
Extrapolation table updated to reflect glycolytic-only FIS1 elevation finding

Pages unverifiable (closed-access):

doi:10.1083/jcb.201007152 (Otera 2010) — tagged no-fulltext-access; claims corroborated by Loson 2013 (verified)

Downstream pages to check:

mff — may cite Otera 2010 for dominance claim; check for correct attribution
drp1 — FIS1-DRP1 interaction description may need verification
mitochondrial-dynamics — overview claims about FIS1 role

[2026-05-06] verify — molecules/proteins/mid49.md (R25 Tier B)

Pages verified: 1 (mid49.md — partial scope; Osellame 2016 and Atkins 2016 remain closed-access)

Sources checked:

doi:10.1038/embor.2011.54 (Palmer 2011, EMBO Reports) — PDF downloaded and read end-to-end; verified
doi:10.1038/emboj.2011.198 (Zhao 2011, EMBO J) — PDF downloaded and read end-to-end; verified
doi:10.1091/mbc.e12-10-0721 (Loson 2013, MBC) — PDF read end-to-end; verified
doi:10.1074/jbc.m113.479873 (Palmer 2013, J Biol Chem) — PDF downloaded and read end-to-end; verified
doi:10.1242/jcs.185165 (Osellame 2016, J Cell Sci) — not_oa (closed-access); no-fulltext-access
doi:10.1042/cs20160030 (Atkins 2016, Clin Sci) — not_oa (closed-access); no-fulltext-access Canonical IDs (UniProt Q96C03, NCBI Gene 125170, HGNC 17920, Ensembl ENSG00000177427) confirmed via UniProt REST API 2026-05-06. Druggability tier confirmed via Open Targets API (ENSG00000177427) — tier 3 (structure with ligand only). COXPD49 clinical description confirmed via UniProt DI-05914 (PMID 29361167).

Corrections:

TOPOLOGY — CRITICAL CORRECTION. Wiki described MID49 as “tail-anchored” with “C-terminal transmembrane helix” and “N-terminus facing the cytoplasm.” This is inverted. Palmer 2011 (Fig 1E–F, proteinase K protection, Na2CO3 extraction, truncation constructs) establishes MID49 is N-terminally anchored: TM helix at aa 23–43 (UniProt confirmed), N-terminus (aa 1–22) in the IMS, cytoplasmic domain is aa 44–454. “Tail-anchored” terminology removed throughout; corrected to “N-terminally anchored.”
PALMER 2013 CELL TYPE — CORRECTED. Wiki footnote stated “MFF/FIS1 DKO HEK293” as the cell type. Palmer 2013 used Mfn-DKO MEFs (mitofusin double-knockout MEFs) to visualize fission events, plus HeLa and 293T for other experiments; Mff/Fis1 independence was shown by siRNA knockdown, not by using DKO cells of those genes. Footnote and body corrected.
COXPD49 CLINICAL DESCRIPTION — CORRECTED. Wiki listed “lactic acidosis” as a clinical feature; this is not in UniProt DI-05914. Correct features per UniProt: progressive muscle weakness, intermittent muscle pain, exercise intolerance, elevated serum creatine kinase, deficiencies of multiple respiratory chain enzymes. Corrected to match verified source.
DRP1 DOMAIN CONTACT — QUALIFIED. Wiki claimed “MID49’s Mab-21 domain contacts the DRP1 GTPase domain.” Palmer 2011 shows Y2H and co-IP confirm direct MiD49–Drp1 interaction, but the specific Drp1 domain contacted is not identified in these papers. Claim softened; “GTPase domain” specificity removed.
DRUGGABILITY — UPDATED. Previous entry had druggability-tier: null due to API 5xx error. Re-queried 2026-05-06: tier 3 (Structure with Ligand is the only positive SM tractability flag). Frontmatter updated; limitations section updated.
PALMER 2013 MITOCHONDRIA-SPECIFICITY — ADDED. Palmer 2013 key finding (MiD49/MiD51 are mitochondria-specific, unlike Fis1/Mff which also target peroxisomes) was understated in wiki; added to body and footnote.
FOOTNOTE METADATA — COMPLETED. Loson 2013 and Palmer 2013 footnotes now include volume/pages (Mol Biol Cell 24(5):659–667; J Biol Chem 288(38):27584–27593). Zhao 2011 footnote corrected to note it primarily characterizes MIEF1 (MiD51), not MiD49.
OSELLAME 2016 AND ATKINS 2016 — FLAGGED UNVERIFIED. Both are closed-access; footnotes updated with NOT VERIFIED warning and no-fulltext-access tags.

Unverifiable claims (closed-access):

Osellame 2016: CRISPR cooperative/independent roles; ER-contact-site specificity; MFF dominance quantification
Atkins 2016: aging-relevant tissue context claims attributed to this review

Downstream pages to check (main agent handles propagation):

mid51 — may contain mirrored topology errors (“tail-anchored”); check
mitochondrial-dynamics-pathway — may reference MFF/FIS1 DKO cell type claim from Palmer 2013; check
drp1 — may cite MID49 as “tail-anchored”; check

[2026-05-06] verify — molecules/proteins/mff.md (R25 Tier B)

Pages verified: 1 (mff.md — partial scope; two sources remain closed-access)

Sources checked:

doi:10.1126/science.aab4138 (Toyama 2016, Science) — green OA; DOI lookup failed; verified via PMC full text (PMC4852862)
doi:10.1091/mbc.e07-12-1287 (Gandre-Babbe 2008, MBC) — PDF read end-to-end; verified
doi:10.1091/mbc.E12-10-0721 (Loson 2013, MBC) — PDF read end-to-end; verified
doi:10.1091/mbc.E15-08-0591 (Liu & Chan 2015, MBC) — PDF read end-to-end; verified
doi:10.1083/jcb.201007152 (Otera 2010, JCB) — not_oa (closed-access); no-fulltext-access
doi:10.1111/cge.13740 (EMPF2 2020, Clin Genet) — not_oa (closed-access); no-fulltext-access Canonical IDs (UniProt Q9GZY8, NCBI Gene 56947, HGNC 24858, Ensembl ENSG00000168772, TM aa 323–340, CC aa 291–322, MW 38,465 Da) confirmed via UniProt REST API.

Corrections:

AMPK phosphosites — CORRECTED. Wiki stated “Ser146 is the best-characterized” AMPK phosphosite on MFF. Toyama 2016 identifies NO Ser146 site. The two validated AMPK phosphorylation sites are Ser155 and Ser172 (human MFF). The phosphonull double mutant is SA2 (Ser155Ala/Ser172Ala); phosphomimetics are SD2 (Ser155Asp/Ser172Asp) and SE2 (Ser155Glu/Ser172Glu). Ser172 mutation alone attenuates most in-vitro phosphorylation; both sites must be mutated to fully ablate it.
ETC inhibitors used — CORRECTED. Wiki claimed AMPK-null cells “fail to fragment” under “antimycin A, oligomycin, CCCP.” Toyama 2016 uses rotenone and antimycin A as the primary ETC inhibitors; CCCP-induced fragmentation operates via a distinct OPA1-cleavage pathway and is NOT the AMPK-dependent mechanism. “Oligomycin” is not used in this context in the paper. Corrected body text accordingly.
Drosophila NMJ — REMOVED. Wiki claimed in-vivo validation was in “Drosophila neuromuscular junction.” Toyama 2016 contains no Drosophila experiments. The in-vivo work uses mouse cortical layer 2/3 pyramidal neurons via in utero electroporation. Body text and footnote corrected.
Mitophagy flux claim — QUALIFIED. Wiki stated “phosphorylation-null MFF mutants block energy-stress-induced fragmentation and reduce mitophagy flux.” Toyama 2016 demonstrates the fragmentation block but does NOT directly measure mitophagy flux. The mitophagy link is inferential. Claim qualified and no-mechanism tag added.
Gandre-Babbe 2008 cell line — CORRECTED. Wiki said MFF identification used “only one cell line (COS-7).” The primary RNAi experiments were in HeLa cells; COS-7 was used for the initial Drosophila screen. Main phenotype quantification: 94 ± 5% cells with elongated mitochondria after siRNA (n=350–400 cells, 4 experiments). Added to body.
Gandre-Babbe overexpression phenotype — CORRECTED. Wiki implied overexpression caused fragmentation (“gain-of-function phenotype”). The paper explicitly shows overexpression does NOT cause fission (6.0 ± 3.8% fragmented in Mff-transfected vs 7.7 ± 1.7% untransfected controls). Corrected body and footnote.
Loson 2013 redundancy claim — CORRECTED. Wiki stated “knockdown of MFF + MID49 + MID51 together produced a more complete elongation than any single knockdown.” Loson 2013 shows MiD49 or MiD51 single knockdown each independently cause elongation; simultaneous double knockdown of MiD49+MiD51 is not more severe than either alone. The more-complete elongation comes from combining Fis1+Mff null (Fis1/Mff-double-null MEFs). Body and footnote corrected.
UniProt phosphosite list — CORRECTED. Removed Ser155 and Ser172 from the “non-AMPK” phosphosite list (these are the validated AMPK sites from Toyama 2016); retained remaining annotated sites (Ser157, 202, 229, 233, 295; Thr115, 200) as sites with unknown kinases.

Unverifiable claims (closed-access):

Otera 2010: MFF as “primary DRP1 receptor”; FIS1 vs MFF siRNA effect-size comparison in HeLa/HCT116 — no-fulltext-access; claims flagged in body
EMPF2 2020: EMPF2 clinical phenotype details, n=1 proband fibroblast data — no-fulltext-access; claims flagged in body

Downstream pages to check (main agent handles propagation):

toyama-2016-ampk-mff-fission — study page likely states Drosophila NMJ and Ser146; needs same corrections
ampk — may reference MFF phosphorylation; Ser146 claim should be corrected to Ser155/Ser172 if present
mitochondrial-dynamics — may reference Gandre-Babbe 2008 overexpression phenotype or Loson 2013 redundancy claims; check
mitophagy — may state that MFF phosphonull mutants reduce mitophagy flux; should be qualified

[2026-05-06] verify — molecules/proteins/yme1l.md (R25 Tier B)

Pages verified: 1 (yme1l.md — partial scope; Wai 2015, Anand 2014, MacVicar 2016, and Leonhard 1996 are closed-access; Elancheliyan 2024 PDF download failed despite gold OA)

Sources checked:

doi:10.1091/mbc.e11-08-0674 (Stiburek 2012, MBC) — PDF downloaded from PMC; read end-to-end; verified
doi:10.7554/elife.16078 (Hartmann 2016, eLife) — PDF downloaded from PMC; read end-to-end; verified
doi:10.1083/jcb.201308006 (Anand 2014, JCB) — not_oa (closed-access); abstract from Crossref; no-fulltext-access
doi:10.1126/science.aad0116 (Wai 2015, Science) — not_oa (closed-access); metadata only; no-fulltext-access
doi:10.1242/jcs.159186 (MacVicar 2016, JCS) — not_oa (closed-access); abstract from Crossref; no-fulltext-access
doi:10.1002/j.1460-2075.1996.tb00796.x (Leonhard 1996, EMBO J) — not_oa (closed-access); DOI verified via Crossref (tb00796.x = correct Leonhard AAA protease paper; tb00819.x cited in seeder brief is an EMBO J frontmatter page); no-fulltext-access
doi:10.1016/j.celrep.2024.115038 (Elancheliyan 2024, Cell Reports) — gold OA but Elsevier redirect blocked DOI lookup (two retries failed); no-fulltext-access Canonical IDs (UniProt Q96TA2, NCBI Gene 10730, HGNC 12843, Ensembl ENSG00000136758) confirmed via UniProt REST + NCBI Gene + HGNC REST APIs.

Corrections:

HGNC ID — CRITICAL CORRECTION. hgnc: 17175 → hgnc: 12843. Confirmed via UniProt Q96TA2 cross-references (HGNC:12843) and HGNC REST API (HGNC:12843 = YME1L1). 17175 is wrong.
OPA1 CLEAVAGE SITE ASSIGNMENT — CRITICAL CORRECTION. Wiki stated “YME1L1 cleaves at S1 site” and “OMA1 cleaves at S2 site.” This is reversed. Hartmann 2016 Fig 3C schematic (read from PDF) explicitly labels: YME1L1 cleaves OPA1 at S2, OMA1 cleaves OPA1 at S1. The Anand 2014 paper (which defines S1/S2) is closed-access, but the Hartmann 2016 paper cites Anand 2014 as the primary definition source and uses the same labeling throughout. Corrected in body, extrapolation table, function section, and disease association table.
MUTATION DESIGNATION — CRITICAL CORRECTION. Wiki stated “R206W, disrupting MPP processing.” Hartmann 2016 PDF confirms the mutation is p.R149W (c.616C>T; NM_014263; arginine 149 → tryptophan in the MTS). R206W appears nowhere in the paper.
FIRST AUTHOR ATTRIBUTION — CORRECTED. Footnote [^sprenger2016] said “Sprenger HG et al.” Sprenger HG is a co-author on Hartmann 2016 but the first author is Hartmann B. All references renamed to [^hartmann2016] and footnote definition corrected.
STIBUREK 2012 MODEL SYSTEM — CORRECTED. Footnote said “YME1L siRNA.” Paper uses stable shRNA knockdown (pGIPZ shRNAmir constructs; selected with puromycin over 3 weeks); not transient siRNA.
STIBUREK 2012 RESPIRATION — CORRECTED. Wiki body said “complex I/IV respiration” impaired. Stiburek 2012 (Fig 6B) shows complex I-specific (CI/CII ratio) respiration diminished ~60%; complex IV activity (CIV/CII, CI/CIV ratios) is unaffected. Complex IV claim removed; quantitative reduction (~60% CI) added.
STIBUREK 2012 APOPTOSIS CLAIM — CLARIFIED. Wiki said “altered MOMP threshold.” Paper specifically shows impaired apoptotic resistance: increased PARP cleavage after staurosporine (2 µM) or H₂O₂ (200 µM) treatment in YME1L KD cells. Mechanism (MOMP threshold) is an inference; corrected to what the paper actually shows.
PRELID1 unsourced RESOLVED. PRELID1 accumulation in YME1L1-deficient patient fibroblasts confirmed in Hartmann 2016 Fig 3B; primary source for constitutive PRELID1 degradation by YME1L1 is Potting et al. 2013 (doi:10.1016/j.cmet.2013.07.008). unsourced tag removed; PRELID1 body text updated; note added that cardiolipin framing needs verification against Potting 2013 (paper establishes phosphatidic acid transport by PRELID1, not cardiolipin directly).
HARTMANN 2016 DISEASE DETAIL — ENRICHED. Added: one consanguineous Saudi Arabian family; 4 affected children (II.5, II.8, II.9, II.11); c.616C>T mutation; full clinical spectrum (intellectual disability, motor developmental delay, optic nerve atrophy, hearing impairment, leukoencephalopathy); YME1L1^R149W^ is hypomorphic (explains survival vs embryonic lethality of complete KO); i-AAA complex assembly (~2 MDa) intact.
LEONHARD 1996 DOI — CONFIRMED CORRECT. Wiki DOI tb00796.x is correct (Crossref confirms this is the Leonhard et al. AAA protease paper). The seeder brief alternative tb00819.x resolves to “Frontmatter” — a wrong DOI. No change needed.

Unverifiable claims (closed-access or failed download):

Wai 2015 (Science): cardiac-specific KO → DCM phenotype; OMA1 double-KO rescue — no-fulltext-access
Anand 2014 (JCB): precise S1/S2 site definition; OMA1-mediated YME1L1 degradation kinetics — no-fulltext-access (S2=YME1L1 confirmed indirectly via Hartmann 2016)
MacVicar 2016 (JCS): OPA1 processing review details — no-fulltext-access
Elancheliyan 2024 (Cell Reports): OCIAD1/prohibitin/TIMM17A axis — PDF download failed despite gold OA; Elsevier redirect blocker; recommend retry or PMC lookup

Downstream pages to check (main agent handles propagation):

oma1 — may document “cleaves S2” (should be S1) or reference “S1=YME1L1”; check and correct
opa1 — may document S1/S2 site assignments; same correction needed
mitochondrial-dynamics-pathway — may cite YME1L1 S1 cleavage; correct to S2
heart-failure — may cite Wai 2015 via YME1L1; verify claim accuracy

[2026-05-06] verify — molecules/proteins/il-18.md (R25 Tier B)

Pages verified: 1 (il-18.md — R25 Tier B; partial scope due to two failed downloads)

Sources checked:

doi:10.1038/378088a0 (Okamura 1995, Nature) — PDF read end-to-end; verified
doi:10.1038/386619a0 (Ghayur 1997, Nature) — PDF read end-to-end; verified
doi:10.1126/science.275.5297.206 (Gu 1997, Science) — PDF read end-to-end; verified
doi:10.1016/s1074-7613(00)80013-8 (Novick 1999, Immunity) — PDF read end-to-end; verified
doi:10.1111/imr.12616 (Kaplanski 2018, Immunol Rev) — PDF read end-to-end; verified
doi:10.1136/annrheumdis-2017-212608 (Gabay 2018, Ann Rheum Dis) — PDF read end-to-end; verified
doi:10.1167/tvst.9.13.27 (Weaver 2020, TVST) — PDF read end-to-end; verified
doi:10.1111/imr.12621 (Dinarello 2018, Immunol Rev) — DOI lookup failed (green OA, PMC mirror fetch failed); no-fulltext-access
doi:10.1016/j.exger.2019.01.020 (Opstad 2019, Exp Gerontol) — DOI lookup failed (green OA, fetch failed); no-fulltext-access Canonical IDs (UniProt Q14116, NCBI Gene 3606, HGNC 5986, Ensembl ENSG00000150782) confirmed via UniProt REST API.

Corrections:

Caspase-1 cleavage site — CORRECTED. “Asp36/Ala37” → “Asp35/Asn36”. Both Gu 1997 (explicit NH₂-terminal sequencing: Asp35–Asn36) and Ghayur 1997 (D35A mutant blocks cleavage) confirm cleavage is at Asp35. The Ala37 residue referenced in the original page was incorrect on both the residue identity (Asn, not Ala) and the position numbering. Corrected in body text and Key interactors section.
Caspase-3 cleavage site — CORRECTED. “Asp71” → “Asp69/Ile70”. Gu 1997 explicitly states CPP32 (caspase-3) cleaves proIGIF at Asp69–Ile70, producing a ~14-kDa fragment with no IFN-γ–inducing activity. Added citation [^gu1997] and added the functional consequence (no IFN-γ activity) which was absent.
IL-18BP molecular weight — CORRECTED. “~21-kDa protein” → “mature protein core of ~20 kDa (164 aa); glycosylated form ~40 kDa on SDS-PAGE.” Novick 1999 explicitly shows the ~40 kDa glycosylated band collapsing to ~20 kDa upon N-glycanase treatment; IL-18BPa has 164 mature aa after 28-aa signal cleavage.
IL-18BP Kd attribution — CORRECTED. Kd ~400 pM was attributed to [^novick1999]. Novick 1999 characterizes the protein but does not report a 400 pM Kd; that value comes from Kaplanski 2018 (review) which states “exceptionally high affinity for IL-18 (400 pM).” Attribution moved to [^kaplanski2018]; novick1999 retained for the structural characterization.
Tadekinig footnote — ENRICHED with exact response rates. Original “~50% response rate by week 3” confirmed accurate (5/10 in 80 mg; 6/12 in 160 mg = 50% each arm). Added n per group breakdown, trial number NCT02398435, and noted 3 SAEs including one possibly drug-related (toxic optic neuropathy, not definitively attributed). Status updated from “pending” to verified.
Weaver 2020 footnote — ENRICHED. Added exact n (27–32 AMD patients, 61 controls), p-value (p=0.0009), AUC (0.73), and adjusted R² (0.1906). The “~19% of IL-18 variance” claim confirmed against the adjusted R² = 0.1906 value in Table 4. Corrected author list to include Cyr B and de Rivero Vaccari JC. Status updated to verified.

Unverifiable claims (failed downloads):

Dinarello 2018 (doi:10.1111/imr.12621): MyD88→IRAK→TRAF6→NF-κB signaling pathway description — no-fulltext-access; tagged in footnote
Opstad 2019 (doi:10.1016/j.exger.2019.01.020): n=300 CAD patients; telomere-length correlation in women — no-fulltext-access; tagged in footnote

Downstream pages to check (main agent handles propagation):

caspase-1 — may document cleavage site as Asp36; should be updated to Asp35/Asn36
nlrp3-inflammasome — may reference IL-18 cleavage site; same correction may be needed
il-1b — parallel page; caspase-1 cleavage site for IL-1β is at Asp116 (different residue; no propagation needed, but worth confirming consistency of notation)

[2026-05-06] verify — molecules/proteins/il-1a.md (R25 Tier B; partial PDF pass)

Pages verified: 1 (il-1a.md — R25 Tier B)

Sources checked:

doi:10.1073/pnas.0905299106 (Orjalo 2009, PNAS) — PDF read end-to-end; verified
doi:10.1073/pnas.0308705101 (Werman 2004, PNAS) — PDF read end-to-end; verified
doi:10.1074/jbc.M306342200 (Buryskova 2004, J Biol Chem) — PDF read end-to-end; verified
doi:10.1084/jem.187.9.1463 (Horai 1998, J Exp Med) — PDF downloaded and read end-to-end; verified
doi:10.1038/315641a0 (March 1985, Nature) — not_oa (closed access); no-fulltext-access
doi:10.4049/jimmunol.1102048 (Rider 2011, J Immunol) — DOI lookup failed (publisher paywall); no-fulltext-access
doi:10.1111/imr.12621 (Dinarello 2017, Immunol Rev) — DOI lookup failed (no OA URL); no-fulltext-access Canonical identity fields (UniProt P01583, NCBI Gene 3552, HGNC 5991, Ensembl ENSG00000115008, NLS 82-86, propeptide 1-112, mature chain 113-271) confirmed against UniProt REST API. druggability-tier:1 confirmed via Open Targets Platform (AB Advanced Clinical = True; bermekimab has reached advanced clinical stage; no approved drug yet for IL-1α specifically).

Corrections:

Werman 2004 footnote — CORRECTED attribution. Original footnote claimed “neither the mature form nor the NLS-deletion mutant retained transcriptional activity.” NLS-deletion mutant (VVATN) result is from Buryskova 2004, not Werman 2004. Werman 2004 only tests pIL-1α (90-fold activation), ppIL-1α (>50-fold), and mIL-1α (inactive) in GAL4 system; no NLS-deletion construct tested in that paper. Footnote rewritten to accurately reflect each paper’s content.
Buryskova 2004 footnote — CORRECTED specificity. Original said “nuclear pro-IL-1α co-precipitates with CBP/p300 and other HAT complexes; its presence increases histone acetyltransferase activity at IL-1α-responsive promoters.” Paper shows interaction with p300 (not CBP), PCAF, Gcn5, and Ada3 via co-IP and GAL4-reporter; it does NOT measure HAT enzymatic activity at IL-1α-responsive promoters. Footnote and body text rewritten accordingly.
Body text (nuclear compartment section) — CORRECTED claim. “binding chromatin and functions as a transcriptional co-activator, particularly driving NF-κB-target gene expression” → “activates transcription 90-fold in GAL4-reporter system; activates NF-κB and AP-1 in IL-1R-independent manner (Werman 2004).” The p300-PCAF-SAGA interaction text now accurately describes what Buryskova 2004 showed (reporter + co-IP) without overclaiming chromatin remodelling at specific loci.
Horai 1998 — REMOVED unsourced claim. Wiki stated “IL-1α-/- mice show impaired contact hypersensitivity responses.” This finding is NOT in Horai 1998, which only tests turpentine-induced fever and glucocorticoid responses. Contact hypersensitivity claim removed; tagged needs-replication pending separate citation.
Horai 1998 — EXPANDED correct findings. Added: IL-1β-/- fever abrogation, mutual induction (IL-1α mRNA >30-fold suppressed in IL-1β-/- brain), glucocorticoid response data, and clarification that the IL-1α/β double-KO lines were also generated.
Orjalo 2009 — ADDED quantitative detail. FACS surface-labeling: 79% senescent vs 20% presenescent. Antibody concentrations: anti-IL-1α 0.6 μg/ml, rIL-1Ra 240 ng/ml, anti-IL-1β 0.8 μg/ml. IRAK1 shRNA knockdown efficiency >90%. C/EBPβ (alongside NF-κB p65) reduction upon IL-1α depletion added to body text. Bleomycin-induced senescence added as primary model (was omitted from original); full list of senescence inducers used in the paper added.
Senescence model specification — CORRECTED. Original described only “Ras-induced and replicatively senescent human fibroblasts.” Paper actually uses 4 models: bleomycin (primary), replicative, RAS^V12-oncogenic, and NaB-induced. Cell line specified as HCA2 primary foreskin fibroblasts.

Unverifiable claims (closed-access / failed download):

Rider 2011: alarmin release mechanism from necrotic cells in sterile peritonitis; timing and cell-recruitment specificity — no-fulltext-access
Dinarello 2017: receptor biology and therapeutic overview claims — no-fulltext-access
March 1985: IL-1α/β gene cloning foundational claim — not_oa; historical background; lower priority

Downstream pages to check (main agent handles propagation):

sasp — may cite Orjalo 2009 for the IL-1α-apex-of-SASP claim; C/EBPβ addition may be relevant
il-1b — Horai 1998 is already cross-referenced there; no direct numerical corrections needed but Horai mutual-induction finding (IL-1α mRNA >30-fold suppressed in IL-1β-/- brain) was not previously documented
chronic-inflammation — may reference IL-1α SASP claims; should be checked against corrected Orjalo 2009 specifics

[2026-05-06] verify — molecules/proteins/il-1r1.md (R25 Tier B; partial PDF pass)

Pages verified: 1 (il-1r1.md — R25 Tier B)

Sources checked:

doi:10.1056/NEJMoa1707914 (Ridker 2017 CANTOS, NEJM) — PDF read end-to-end; verified
doi:10.1074/jbc.270.23.13757 (Greenfeder 1995, J Biol Chem) — PDF read end-to-end; verified
doi:10.1016/s1074-7613(00)80402-1 (Wesche 1997, Immunity) — PDF read end-to-end; verified
doi:10.1126/science.2969618 (Sims 1988, Science) — not_oa; claims flagged in verified-scope
doi:10.1016/j.smim.2013.10.023 (Boraschi & Tagliabue 2013, Semin Immunol) — not_oa; claims flagged in verified-scope
doi:10.1146/annurev.immunol.021908.132612 (Dinarello 2009, Annu Rev Immunol) — not_oa; claims flagged in verified-scope

Corrections:

Beta-bulge/QGEESN attribution to Greenfeder 1995 — REMOVED. Greenfeder 1995 shows IL-1Ra fails to recruit IL-1RAcP but does NOT describe the beta-bulge structural element or the QGEESN sequence swap experiment. The beta-bulge claim is from later crystallographic work. Text corrected to accurately reflect what Greenfeder 1995 shows and the beta-bulge claim re-attributed as needing a primary citation (#gap/needs-citation added).
CANTOS footnote trial description — EXPANDED. Was “canakinumab 150 mg q3m vs placebo” (understated trial design). Corrected to note all three doses tested (50 mg, 150 mg, 300 mg q3m), that 150 mg was the only dose meeting the prespecified significance threshold (HR 0.85, 95% CI 0.74–0.98, P=0.021), and that all-cause mortality was neutral (HR 0.94, P=0.31). Fatal infection adverse event rate added.

Unverifiable claims (closed-access sources):

Sims 1988: IL-1R1 as 569 AA type I single-pass transmembrane glycoprotein; immunoglobulin superfamily classification
Boraschi & Tagliabue 2013: D3 domain role in IL-1RAcP recruitment; Tyr496 phosphorylation → PI3K-p85; six N-glycosylation site count
Dinarello 2009: ligand selectivity table characterisations

Downstream pages to check:

canakinumab — already verified (R12); CANTOS numerics live there; no correction needed from this pass
il-1-signaling — implicit stub seeded in R25; no content yet to propagate
chronic-inflammation — may reference il-1r1 signalling claims; check on next lint pass

[2026-05-06] verify — cell-types/fibroadipogenic-progenitors.md (full PDF pass)

Pages verified: 1 (fibroadipogenic-progenitors.md — R25 Tier A; full PDF pass)

Sources checked:

doi:10.1038/ncb2015 (Joe 2010, Nat Cell Biol) — PDF read end-to-end; verified
doi:10.1038/ncb2014 (Uezumi 2010, Nat Cell Biol) — PDF read end-to-end; verified
doi:10.1016/j.cell.2013.02.053 (Heredia 2013, Cell) — PDF read end-to-end; verified (stale DOI lookup corrected to WD/…)
doi:10.1016/j.celrep.2019.04.074 (Wosczyna 2019, Cell Reports) — CORRECTED DOI from wrong 10.1016/j.celrep.2019.07.073 (was a VISTA protein paper); downloaded fresh; PDF read end-to-end; verified
doi:10.1038/s41418-020-0551-y (Reggio 2020, Cell Death Differ) — PDF downloaded and read end-to-end; verified
doi:10.1038/nm.3869 (Lemos 2015, Nat Med) — not_oa; claims flagged no-fulltext-access

Corrections:

Wosczyna 2019 DOI: 10.1016/j.celrep.2019.07.073 → 10.1016/j.celrep.2019.04.074 (PMID:31091443). Previous DOI resolved to a VISTA immunology paper.
Wosczyna 2019 ablation mechanism: “diphtheria toxin receptor (DTR) system” → “PDGFRα^CreER;R26^DTA system (Cre-dependent DTA expression)”; DTR was never used in this paper.
Wosczyna 2019 timeline: “satellite-cell numbers decline within weeks” → “approximately 45% reduction in MuSC numbers over 9 months under homeostatic conditions”; the weeks timeline is unsupported by the paper.
Wosczyna 2019 unsupported claim removed: “bidirectional cross-talk (satellite cells also signal back to FAPs)” — this is not a finding of the paper.
Heredia 2013 eosinophil model: “IL-5-null” → “ΔdblGATA (GATA-1 high-affinity site deletion)”. IL-5-null is a different knockout; this paper uses ΔdblGATA.
Heredia 2013 rescue: “rescued by IL-4 injection” → “partially rescued by IL-4 complex injection (IL-4 + anti-IL-4 antibody complex, n=6)”; simple IL-4 was not used.
Heredia 2013 FAP isolation: paper uses MACS CD31⁻/CD45⁻/α7-integrin⁻/Sca-1⁺ gate (≥98% pure) not the Joe 2010 Hoechst/PI gate.
Reggio 2020 MAJOR CORRECTION: entire “Senescent FAPs in chronic muscle disease” section removed — paper does NOT describe p16^INK4a, p21, SA-β-gal+ FAP subpopulation, does NOT report SASP with IL-6/IL-8/MMPs, does NOT test ABT-263/navitoclax. All these claims were fabricated by initial extraction. Paper instead reports WNT5a/GSK3/β-catenin axis regulation of FAP adipogenesis.
Reggio 2020: study page slug corrected from “reggio-2020-senescent-fap-wnt-adipogenesis” → “reggio-2020-wnt-gsk3-fap-adipogenesis” in footnote to reflect actual content.
Joe 2010 isolation strategy clarified: FAPs are Sca-1⁺/CD34⁻ gate (not Lin⁻/PDGFRα⁺ as primary gate); PDGFRα+ used for in-situ/transplantation validation; >85% of FAPs are PDGFRα+.
Uezumi 2010 mouse exclusion marker: “CD56” → “SM/C-2.6” (satellite-cell antibody used in this paper); CD56 is a human marker convention, not used in Uezumi 2010 mouse FACS.
Human marker panel: key-markers-human updated to remove CD56 as formally gated marker (not explicitly gated in Uezumi 2010 human panel).
Hallmark connections table updated: chronic-inflammation and cellular-senescence rows corrected to reflect verified evidence.
Therapeutic implications: senolytics bullet corrected to note ABT-263/navitoclax is uncited; GSK3 inhibition (LY2090314) and WNT5a restoration added as the Reggio 2020 supported strategies.

Unverifiable claims remaining:

Lemos 2015 (doi:10.1038/nm.3869) — not_oa; TNF-α/TNFR1/caspase-8 apoptosis-switch mechanism and nilotinib mdx result unverified

Downstream pages to update (main agent):

studies/wosczyna-2019-fap-satellite-cell-niche — DOI field needs correction to 10.1016/j.celrep.2019.04.074
studies/reggio-2020-senescent-fap-wnt-adipogenesis — page title/slug misrepresents content; should be renamed to reflect WNT/GSK3 axis focus; p16/p21/navitoclax claims must be removed
Any entity pages citing [^wosczyna2019] with DTR or “within weeks” language
Any entity pages citing [^heredia2013] with “IL-5-null” language
Any entity pages citing [^reggio2020] for senescent FAP / navitoclax claims

[2026-05-06] verify — tissues/heart.md (full PDF re-pass)

Pages verified: 1 (heart.md — full PDF pass replacing prior training-knowledge-only pass)

Sources checked:

doi:10.1016/j.hfc.2011.08.011 (Strait & Lakatta 2012 Heart Fail Clin) — read via PMC eutils full-text XML (PMC3223374); three quantitative claims corrected
doi:10.1126/science.1164680 (Bergmann 2009 Science) — abstract confirmed via PubMed PMID 19342590; DOI lookup failed; PMC2991140 partial fetch; renewal rates ~1%/25yr and ~0.45%/75yr confirmed
doi:10.1001/jama.285.18.2370 (Go 2001 JAMA) — abstract confirmed via PubMed PMID 11343485; not OA; age-bracket for 0.1% figure corrected; 2.3%/60-69yr figure not in abstract

Corrections:

LV wall thickness claim “~25-30% between ages 25 and 80” — REMOVED: Strait & Lakatta 2012 states LV mass does not increase in healthy aging (decreases in men, unchanged in women); wall thickening is asymmetric/septal only. Replaced with accurate qualitative description + unsourced for any % quantification; BLSA imaging studies should be cited for specific figures.
SAN pacemaker cell absolute count “~10,000–20,000 cells” — NOT in Strait & Lakatta 2012; paper only states “<10% remain by age 75 relative to young-adult count”; absolute count flagged unsourced with reference to Shiraishi 1992 as likely primary source.
Aortic sclerosis prevalence “~25% at 65-74, ~50% at ≥75” attributed to [^straitlakatta2012] — REMOVED from that citation; paper discusses aortic calcification mechanistically without prevalence stats; flagged unsourced; likely primary sources noted (Otto 1999 NEJM / Stewart 1997 JACC).
Go 2001 AF prevalence age bracket corrected: “Age 20–39: ~0.1%” → “Age <55 years: ~0.1%”; abstract states “<55 years” not “20-39”.
Go 2001 intermediate figure “~2.3% at age 60–69” — NOT in abstract; flagged as unverifiable (likely in full-text Table 1, which is closed-access).
Max HR “~1 bpm/year” claim — nuanced; Strait & Lakatta 2012 discusses max HR decline qualitatively but does not state this specific rate; note added.
All three footnotes ([^straitlakatta2012], [^bergmann2009], [^go2001]) updated to document verification method and confirmed/corrected findings.
Banner updated from “⚠️ Partially verified” to completed-pass note.
verified-scope updated with full PDF-read documentation.

Unverifiable claims remaining:

Lakatta & Levy 2003 (doi:10.1161/01.cir.0000048892.83521.58) — closed-access; not verified this pass
Go 2001 full-text table data (2.3% at 60-69, other age-specific bands) — closed-access
Bergmann 2009 exact n — not in abstract; PMC full-text XML body not extracted
SAN absolute cell count (~10,000-20,000) — Shiraishi 1992 not in archive; unsourced maintained
Aortic sclerosis prevalence — Otto 1999/Stewart 1997 not verified; unsourced maintained

Downstream propagation needed:

None required immediately — heart.md is an organ-level synthesis MOC; quantitative claims corrected here are not cited verbatim on downstream pages (atomic claims live on myocardium which is already verified). Main agent should note that any page citing “LV wall thickness ~25-30%” or “~25%/~50% aortic sclerosis” as sourced from Strait & Lakatta 2012 would inherit this correction.

[2026-05-06] verify — cell-types/fibroadipogenic-progenitors.md

Pages verified: 1 (partial scope — Joe 2010/Uezumi 2010/Heredia 2013 cross-checked against training knowledge; PDFs locally available but paths unresolvable without bash; Lemos 2015 not_oa; Wosczyna 2019 DOI-title mismatch; Reggio 2020 pending download)

Sources checked:

doi:10.1038/ncb2015 (Joe 2010 Nat Cell Biol) — locally available per archive; PDF path unresolvable without bash (BUG-5); claims cross-checked against training knowledge; appear consistent
doi:10.1038/ncb2014 (Uezumi 2010 Nat Cell Biol) — locally available per archive; PDF path unresolvable without bash (BUG-5); claims cross-checked against training knowledge; appear consistent
doi:10.1016/j.cell.2013.02.053 (Heredia 2013 Cell) — locally available per archive; PDF path unresolvable without bash (BUG-5); claims cross-checked against training knowledge; appear consistent
doi:10.1038/nm.3869 (Lemos 2015 Nat Med) — not_oa; no-fulltext-access applied to TNF-α/TNFR1/caspase-8 mechanism and nilotinib mdx result
doi:10.1016/j.celrep.2019.07.073 (Wosczyna 2019 Cell Rep) — archive returns DOI-title mismatch (BUG-2; VISTA protein paper returned); claims retained with needs-replication; DOI accuracy unresolvable without WebFetch
doi:10.1038/s41418-020-0551-y (Reggio 2020 Cell Death Differ) — pending download; no-fulltext-access applied to all quantitative claims (senescent FAP markers, navitoclax histology metrics, WNT5a/GSK3/β-catenin mechanism)

Corrections:

Added no-fulltext-access tags to Lemos 2015-sourced claims (TNF-α mechanism body text, nilotinib mdx result, fibrogenic differentiation)
Added no-fulltext-access tags to Reggio 2020-sourced claims (senescent FAP section, WNT5a/adipogenic shift section, therapeutic implications section)
Added needs-replication note inline to Wosczyna 2019 body section (clarifying archive DOI-title mismatch)
Expanded Limitations section with explicit no-fulltext-access entry listing Lemos 2015 and Reggio 2020 unverified claims
Added verification note to Limitations explaining bash-constraint and recommending re-pass with bash access
Updated Reggio 2020 footnote from “pending download” to explicit no-fulltext-access with enumerated unverified claims
Updated Lemos 2015 footnote to explicitly flag no-fulltext-access

Unverifiable claims:

All Lemos 2015 quantitative details (exact n, p-values, TNF-α dose-response, TNFR1/caspase-8 mechanistic detail, nilotinib dose in mdx) — not_oa permanently
All Reggio 2020 quantitative details (n, p16+/SA-β-gal frequencies, navitoclax dose/histology metrics, WNT5a expression changes) — pending download
Wosczyna 2019 DTR-specific quantitative data (n, satellite cell count decline timeline, myofiber size change) — DOI-title mismatch blocks archive access

Downstream propagation needed:

None identified — this is a new atomic page; no downstream pages yet cite these specific quantitative claims
When Reggio 2020 download completes, re-verify WNT5a/GSK3/β-catenin adipogenic-shift claim and senescent FAP SASP profile
When Lemos 2015 access becomes available (institutional access or preprint), verify TNF-α apoptosis switch mechanism and nilotinib mdx quantitative result

[2026-05-06] verify — molecules/proteins/oma1.md (R25 Tier B)

Pages verified: 1 (oma1.md — partial scope; Anand 2014, Alavi 2022, MacVicar 2016 unverifiable)

Sources checked:

doi:10.1083/jcb.200906084 (Ehses 2009, JCB) — PDF downloaded (bronze OA PMC) and read end-to-end; verified
doi:10.1083/jcb.200906083 (Head 2009, JCB) — PDF downloaded (PMC) and read end-to-end; verified
doi:10.1002/embj.201386474 (Baker 2014, EMBO J; seeder cited as “Quiros 2014”) — PDF downloaded (bronze OA) and read end-to-end; verified
doi:10.1074/jbc.M305584200 (Käser 2003, JBC) — PDF downloaded (camoufox) and read end-to-end; verified
doi:10.1126/scitranslmed.aan4935 (Acin-Perez 2018, Sci Transl Med) — PDF downloaded (camoufox) but is supplementary PDF only; main paper content verified via Crossref abstract + Europe PMC
doi:10.1083/jcb.201308006 (Anand 2014, JCB) — not_oa (closed-access); no-fulltext-access
doi:10.1016/j.cmet.2022.08.017 (Ahola 2022, Cell Metab) — PDF download failed repeatedly (camoufox Cloudflare block); verified via Europe PMC abstract; no-fulltext-access
doi:10.1016/j.phrs.2022.106063 (Alavi 2022, Pharmacol Res) — download failed (no OA URL found); no-fulltext-access
doi:10.1242/jcs.159186 (MacVicar 2016, JCS) — not_oa (closed-access); no-fulltext-access Canonical IDs verified via UniProt REST API (Q96E52) and HGNC REST API: UniProt Q96E52 confirmed; NCBI Gene 115209 confirmed; Ensembl ENSG00000162600 confirmed; HGNC 29661 confirmed (seeder used 23858 — wrong).

Corrections:

HGNC ID — CRITICAL CORRECTION. hgnc: 23858 → hgnc: 29661. Confirmed via UniProt Q96E52 cross-references (HGNC:29661) and HGNC REST API (symbol OMA1 = HGNC:29661).
OPA1 CLEAVAGE SITE ASSIGNMENT — CRITICAL CORRECTION (same error pattern as YME1L1 verification). Wiki assigned OMA1→S2 and YME1L→S1. All three primary PDFs (Ehses 2009, Head 2009, Baker 2014) confirm: OMA1 cleaves OPA1 at S1; YME1L cleaves OPA1 at S2. Corrected throughout: OPA1 cleavage table, body text, Key interactors section, intro paragraph.
OMA1 CONSTITUTIVE ACTIVITY — CORRECTED. Wiki described OMA1 as “constitutively inactive.” Baker 2014 demonstrates OMA1 is constitutively active at low basal levels (generates OPA1 forms c and e in resting cells); stress dramatically amplifies activity. “Constitutively inactive” removed throughout; corrected to “constitutively active at low basal levels.”
AUTOCATALYTIC CLEAVAGE DIRECTION — CORRECTED. key-ptms listed “autocatalytic N-terminal cleavage” only. Baker 2014 shows degradation initiates from both N- and C-termini, resulting in complete OMA1 turnover (not a partial processing). Corrected in frontmatter key-ptms and body.
ACTIVATION STIMULI — NUANCED. Wiki said “ATP depletion — synergizes with depolarization.” Baker 2014 does not show strict correlation between ATP depletion and OMA1 activation under all conditions; oligomycin activates partly; combined oligomycin+antimycin A is required for strong activation. Also added H2O2 oxidative stress as verified stimulus. Description rewritten.
ACTIVE SITE TOPOLOGY — CORRECTED. Wiki domain table stated “HExxH motif; faces matrix.” Baker 2014 (Fig 2C topological model, Fig 2B Na2CO3 extraction, Fig 2D TEV accessibility) establishes the C-terminal M48 catalytic domain faces the IMS, not the matrix. Corrected in domain table.
N-TERMINAL PROCESSING MISATTRIBUTION — CORRECTED. Body text stated “m-AAA protease afg3l2 processes the OMA1 N-terminus for activation competence [^kaser2003].” This claim is not in Kaser 2003 (yeast paper; says nothing about mammalian OMA1 maturation). Baker 2014 shows the mature ~43 kDa OMA1 form arises by N-terminal cleavage at Ala-140 upon import, but the responsible protease is unknown. Text and kaser2003 footnote corrected; afg3l2 interaction note in Key interactors corrected.
FOOTNOTE FIRST AUTHOR — CORRECTED. [^quiros2014] credited “Quirós PM et al.” The actual paper (doi:10.1002/embj.201386474) has first author Baker MJ; Quirós is a co-author not first author. New [^baker2014] footnote added; [^quiros2014] retained as legacy key pointing to same DOI with correction note.
ACIN-PEREZ 2018 CARDIAC MODELS — CORRECTED. Wiki stated three models were “pressure overload, ischemia-reperfusion, doxorubicin.” Crossref abstract confirms “three murine models of heart failure with diverse etiologies”; supplementary figures show models are isoproterenol (tachycardiomyopathy), angiotensin II/pressure overload (HFpEF), and high-fat diet (metabolic cardiomyopathy). Body and footnote corrected.
DISULFIDE Cys407-Cys465 — TAGGED UNSOURCED. This claim appeared in key-ptms and domain table; not found in any of the primary PDFs reviewed. Tagged unsourced; left in place with warning pending source identification.
KORWITZ 2016 DOI MISMATCH (BUG-2) — DOCUMENTED IN LIMITATIONS. Confirmed: real Korwitz 2016 paper is doi:10.1083/jcb.201507022 (PMID 26783299), a neurodegeneration paper (“Loss of OMA1 delays neurodegeneration by preventing stress-induced OPA1 processing in mitochondria”). This is NOT the cardiac protection paper — that remains Acin-Perez 2018. Limitation note updated accordingly.
Ehses 2009 footnote S2→S1 corrected. Head 2009 footnote updated with quantitative detail (56% inhibition, SD=12%, n=5 experiments). Kaser 2003 footnote corrected to remove fabricated mammalian-processing claim.

Unverifiable claims (closed-access / failed download):

Anand 2014 (JCB, 10.1083/jcb.201308006): OMA1/YME1L competitive axis stoichiometry, double-KO constitutively-fused phenotype — no-fulltext-access (claim directionally supported by Baker 2014 which cites Anand 2014)
Ahola 2022 (Cell Metab): ferroptosis mechanism details (GPX4 levels, lipid peroxidation quantification, full pathway) — PDF unavailable; abstract verified via Europe PMC
Alavi 2022 (Pharmacol Res): druggability discussion details — no-fulltext-access
MacVicar 2016 (J Cell Sci): OMA1 aging/neurodegeneration review claims — closed-access no-fulltext-access

Downstream propagation needed (main agent):

opa1 — likely documents S1/S2 assignments; given two prior verifications (YME1L1, OMA1) found this inverted, opa1 must be checked and corrected if still wrong
yme1l — already corrected in prior pass; cross-check OMA1 S1 reference is consistent with the yme1l.md correction
mitochondrial-dynamics pathway page — may reference OMA1 S2 cleavage; check
Any page citing Korwitz 2016 (doi:10.1083/jcb.201510104) for cardiac OMA1 protection — that DOI is wrong; should cite Acin-Perez 2018

Final verified state: verified: true (partial scope — see verified-scope field in frontmatter)

[2026-05-06] R25 — Atomic-content backlog (Tier A + Tier B; 18 pages)

Round 25 cleared the highest-priority items from the R25+ backlog (R24-skipped + 8+-inbound unblocks + R12/R13 implicit-stub clusters). Three sub-batches; all 18 pages seeded verified: false with ⚠️ banners.

Tier A (5 pages, R24-skipped + highest-unblock):

cell-types/fibroadipogenic-progenitors.md — FAPs; PDGFRα⁺/Sca1⁺ muscle interstitial mesenchymal progenitor; 9+ inbound. Closes skeletal-muscle key-cell-type cross-link. (Joe 2010, Uezumi 2010, Lemos 2015, Heredia 2013, Wosczyna 2019, Reggio 2020.)
pathways/heat-shock-response.md — HSR pathway; closes the hsf1 (R24a) → hsp70 (R13) axis flagged in ROADMAP. KEGG null (no canonical HSR entry); Reactome R-HSA-3371556. (Anckar 2011, Akerfelt 2010, Hsu 2003, Beere 2000, Locke 1996.)
molecules/compounds/elamipretide.md — SS-31 cardiolipin stabilizer; biologic peptide; PubChem CID 11764719 (corrected from brief’s 11765011 — different compound). Phase 3 in Barth syndrome (TAZPOWER) + dry AMD; human-evidence-level: limited-negative. (Szeto 2014, Thompson 2021, Marcinek 2019, Sabbah 2016, López-Otín 2023.)
pathways/ras-mapk.md — RAS-RAF-MEK-ERK; KEGG hsa04010, Reactome R-HSA-5673001. Druggability tier 1 (KRAS-G12C inhibitors sotorasib/adagrasib FDA-approved; trametinib MEKi). (Serrano 1997 OIS, Slack 2015 Drosophila trametinib lifespan, Roskoski 2012, Lake 2016.)
tissues/heart.md — organ-level synthesis-MOC; companion to verified myocardium. type: tissue (organ-level convention pending schema formalization). (Strait & Lakatta 2012, Lakatta & Levy 2003, Bergmann 2009, Go 2001 AF prevalence.)

Tier B IL-1 + GPCR cluster (7 pages):

molecules/proteins/gpr43.md — FFAR2; UniProt O15552. SCFA receptor; Maslowski 2009 KO colitis; Smith 2013 Treg/HDAC.
molecules/proteins/gpr41.md — FFAR3; UniProt O14843. Sympathetic ganglia + L-cells; Kimura 2011 SNS axis (locally available PDF).
molecules/proteins/gpr109a.md — HCAR2; UniProt Q8TDS4. Niacin / BHB / butyrate receptor; Tunaru 2003, Singh 2014 Treg, Moutinho 2022 microglia/AD.
molecules/proteins/il-1r1.md — UniProt P14778. Primary IL-1 receptor; Sims 1988; CANTOS reuse.
molecules/proteins/il-1a.md — UniProt P01583. Dual nuclear/alarmin function; Orjalo 2009 SASP-apex mechanism (IL-1α blockade abrogates IL-6/IL-8); bermekimab.
molecules/proteins/il-1ra.md — IL1RN; UniProt P18510. Anakinra = recombinant. Eisenberg/Hannum 1990 concurrent cloning; Larsen 2007 NEJM T2D RCT.
molecules/proteins/il-18.md — UniProt Q14116. Caspase-1-cleaved alongside IL-1β; tadekinig alfa IL-18BP; Okamura 1995 cloning.

Tier B mitochondrial dynamics receptor + protease cluster (6 pages):

molecules/proteins/mff.md — UniProt Q9GZY8. Primary mammalian DRP1 OMM receptor; AMPK-Ser146 phospho-substrate (Toyama 2016 Science). Toyama DOI corrected by seeder (aab3009 → aab4138).
molecules/proteins/fis1.md — UniProt Q9Y3D6. Accessory mammalian DRP1 receptor (vs primary in yeast); P110 inhibitor of FIS1-DRP1.
molecules/proteins/mid49.md — MIEF2; UniProt Q96C03. Critical canonical-DB corrections by seeder via UniProt REST: NCBI Gene 125170 (not 51024 / 117177 as in brief); Ensembl ENSG00000177427 (not ENSG00000108829).
molecules/proteins/mid51.md — MIEF1; UniProt Q9NQG6. ADP-binding allosteric module (Loson 2014). HGNC corrected by seeder (25979 vs brief’s 21931).
molecules/proteins/oma1.md — UniProt Q96E52. Stress-activated zinc metalloprotease; OPA1 S2-site cleavage. DOI mismatch caught: brief’s “Korwitz 2016” DOI 10.1083/jcb.201510104 resolves to a Numb/p53 paper; Acin-Perez 2018 substituted as cardiac OMA1 reference.
molecules/proteins/yme1l.md — UniProt Q96TA2. i-AAA constitutive protease; Wai 2015 Science cardiac KO. Two DOIs corrected by seeder (Stiburek 2012, MacVicar 2016).

Cross-cutting propagation (implicit-stub cleanup):

opa1 — replaced OMA1/YME1L “implicit stub” notes with explicit cross-links; removed gap entry.
drp1 — stripped “implicit stub” suffixes from MID49/MID51 entries.
mitochondrial-dynamics pathway MOC — replaced 6 “implicit stub” suffixes with brief drafted-R25 annotations.
grb2 — replaced ras-mapk implicit-stub note with confirmed cross-link.
myocardium — heart note updated from implicit-stub to drafted-R25.
il-1b — replaced “Implicit stubs created by this page” block with R24/R25 family-completion cross-link summary.
nlrp3-inflammasome — updated atomic-protein-page status note; added R25 IL-1a/IL-1ra/IL-1r1 family-completion cross-links.
caspase-1 — removed #gap/needs-stub for IL-18; updated to drafted R25.

DOI / canonical-DB corrections caught at seed time (8+):

PubChem CID 11765011 → 11764719 (elamipretide; brief’s CID resolved to a different C33H28F3N5O5 compound).
Toyama 2016 Science DOI aab3009 → aab4138 (MFF-AMPK paper).
NCBI Gene 51024/117177 → 125170 (MID49/MIEF2).
Ensembl ENSG00000108829 → ENSG00000177427 (MID49).
HGNC 21931 → 25979 (MID51).
Werman 2004 PNAS DOI 0308686100 → 0308705101 (IL-1α NLS paper).
Stiburek 2012 DOI E11-08-0712 → e11-08-0674 (YME1L paper, different title).
MacVicar 2016 DOI 183343 → 159186 (OPA1 review).
Korwitz 2016 DOI 10.1083/jcb.201510104 (cardiac OMA1) — resolves to unrelated Numb/p53 paper; Acin-Perez 2018 substituted (10.1126/scitranslmed.aan4935).
Dinarello 2018 IL-1 review DOI 10.1002/eji.201746939 — resolves to CD1-T-cell paper; replaced with 10.1111/imr.12621.
Boraschi 2013 DOI 10.1111/imr.12015 — resolves to NK cell paper; replaced with 10.1016/j.smim.2013.10.023.
Novick 1999 IL-18BP DOI 10.1016/s1074-7613(00)80009-5 — not found; corrected to 10.1016/s1074-7613(00)80013-8.
Calderwood 2009 TIBS DOI 10.1016/j.tibs.2009.10.003 — resolves to mucin paper; replaced with 10.1159/000225957.
Morimoto 2008 CSHSQB DOI 10.1101/sqb.2007.72.041 — resolves to circadian paper; replaced.
PROGRESS-HF (elamipretide HF) DOI 10.1002/ejhf.876 — resolves to ivabradine paper; citation dropped.
Offermanns BHB-GPR109A 2014 Nat Med DOI 10.1038/nm.3700 — resolves to tau/asparagine endopeptidase paper; Youm 2015 (10.1038/nm.3804) substituted with caveat that it documents NLRP3 inhibition rather than GPR109A-dependent effect.

Schema escalations surfaced for R26+:

Pathway page druggability-tier ambiguity (heat-shock-response seeder): tier should reflect aging-indication druggability or maximum-druggability of any node? mTOR uses tier 1 (rapamycin = aging drug); HSR also assigned tier 1 (HSP90 inhibitors = oncology drugs, not aging). Recommend druggability-tier-aging-context: field or prose convention to distinguish.
Pathway page verified: field not in CLAUDE.md schema explicitly (mtor.md uses it; HSR follows that precedent). Recommend formalizing in CLAUDE.md type: pathway block.
mr-causal-evidence: partial vs not-tested for GWAS-hit-but-no-MR-instrument situations (il-1ra seeder). Threshold not defined; clarification needed.
cardiolipin-stabilizer intervention class should be promoted from intervention-classes.md roadmap to the main inventory with a Dataview block (elamipretide is the canonical first compound).
MID49/MID51 canonical-DB drift — the MIEF1/MIEF2 family had at least three wrong IDs in my prior R12+R13 implicit-stub references. Recommend a lint-pass UniProt REST cross-check against all in-scope protein pages with verified: false.

Verification status: All 18 pages ship verified: false with ⚠️ banner; await wiki-verifier sweep. DOI sanity caught at seed time (16 corrections).

Cell-Ontology / canonical-ID gaps:

FAPs have no dedicated CL term (CL:0002320 is too broad); cell-ontology-id: null + #gap/needs-canonical-id.
HSR has no standalone KEGG entry (hsa04141 is ER-UPR); kegg: null.

Open Targets API failures during seeding (5 pages): druggability-tier left null on mff/fis1/mid49/mid51/oma1/yme1l due to GraphQL 5xx errors. Re-query needed before lint pass.

[2026-05-06] verify — molecules/proteins/gpr109a.md

Pages verified: 1

molecules/proteins/gpr109a.md — 4 corrections; verified: true (partial scope — Tunaru 2003 and Singh 2014 verified against primary PDFs; 6 other sources not read this pass)

Sources verified against PDFs:

doi:10.1038/nm824 (Tunaru 2003, Nature Medicine) — PDF at — read in full (4 pages); confirmed Gαi coupling, PUMA-G-KO complete abolition of niacin anti-lipolytic effect, EC50 values
doi:10.1016/j.immuni.2013.12.007 (Singh 2014, Immunity) — PDF at — read in full (10 pages); confirmed Treg quantification (~25.8% WT vs ~14.4% Niacr1−/− colonic LP), IL-18 colonocyte requirement, DSS colitis lethality in KO, ApcMin/+ carcinogenesis data

Corrections made (4):

Tunaru 2003 footnote receptor name: “HM74A (human)” → “HM74 (human)” — Tunaru 2003 used HM74 (= GPR109B, low-to-mid affinity); HM74A (= GPR109A high-affinity) is the Wise 2003 distinction; conflating the two misattributes the deorphanization to the wrong receptor
Tunaru 2003 footnote — added verified EC50 values: ~3 µM for mouse PUMA-G, ~1 µM for human HM74 (Ca²⁺ mobilization assay with Gα15); these differ from the ~0.1 µM value in the ligand table, which is Wise 2003’s HM74A figure
Tunaru 2003 footnote local PDF path: (stale local path) → (stale mount point in original extraction)
Ligand table niacin EC50: expanded from bare “~0.1 µM” to disambiguate sources — “~0.1 µM (HM74A, cAMP assay) [^wise2003]; ~1 µM (human HM74) / ~3 µM (mouse PUMA-G) in Ca²⁺ mobilization assay [
Singh 2014 footnote — added quantitative Treg data (Foxp3+ ~25.8% WT vs ~14.4% KO, p<0.002), DSS colitis lethality detail, ApcMin/+ polyp data, and mouse strain (C57BL/6 background, 5× backcross)

Sources unverified this pass (archive status: locally available but not prioritized, or pending):

doi:10.1074/jbc.M210695200 (Wise 2003) — pending; EC50 ~0.1 µM for HM74A claim retained unverified
doi:10.1074/jbc.C500213200 (Taggart 2005) — pending; BHB EC50 ~0.7 mM retained unverified
doi:10.1158/0008-5472.CAN-08-4466 (Thangaraju 2009) — pending
doi:10.1038/nm.3804 (Youm 2015) — pending; NLRP3-independent BHB mechanism caveat structurally intact in body text
doi:10.1126/scitranslmed.abl7634 (Moutinho 2022) — pending
doi:10.1056/NEJMoa1107579 (AIM-HIGH 2011) — pending

Downstream pages to check (main agent):

None identified — gpr109a.md is not cited by any other wiki page via footnote (no study page exists for these papers); no propagation required this pass

[2026-05-06] verify — molecules/proteins/il-1ra.md (R25 Tier B; partial PDF pass)

Pages verified: 1 (il-1ra.md — R25 Tier B; verified: true with scope)

Sources checked:

doi:10.1056/NEJMoa065213 (Larsen 2007, NEJM) — PDF read end-to-end; verified
doi:10.1084/jem.191.2.313 (Horai 2000, J Exp Med) — PDF read end-to-end; verified (downloaded from PMC during this pass)
doi:10.1016/j.amjcard.2013.08.047 (Van Tassell 2014, Am J Cardiol) — abstract/metadata only via Europe PMC + Crossref; full PDF download failed; partial verification
UniProt P18510 / NCBI Gene 3557 / HGNC 6000 / Ensembl ENSG00000136689 — confirmed via HGNC REST API
doi:10.1038/343341a0 (Eisenberg 1990), doi:10.1038/343336a0 (Hannum 1990), doi:10.1146/annurev.immunol.16.1.27 (Arend 1998), doi:10.1002/1529-0131(199812)41:12<2196::AID-ART15>3.0.CO;2-2 (Bresnihan 1998) — not_oa; not verified

Corrections:

Larsen 2007 HbA1c p-value: “p<0.05” → “P=0.03 (95% CI 0.01–0.90)”. Paper reports exact p=0.03.
Larsen 2007 n description: added arm-level breakdown (34 anakinra / 35 placebo; 67 completed). Footnote expanded to include CI, C-peptide p=0.05, proinsulin/insulin p=0.005, and note that insulin sensitivity was unchanged (P=0.58).
Horai 2000 arthritis onset: “by 8 weeks of age” → corrected to reflect actual paper data: detectable from 5 weeks, >80% before 8 weeks, 100% by 13 weeks.
Horai 2000 “disease requires T cells” REMOVED — this claim is not supported by Horai 2000. The paper shows T/B-cell activation and autoantibody production but does not demonstrate T-cell requirement via depletion or transfer experiments.
Horai 2000 autoantibody types specified: anti-IgG (rheumatoid factor), anti-type-II-collagen, anti-dsDNA — per paper’s Table I and Fig. 3 data.
Horai 2000 strain detail expanded: C57BL/6 incidence 0/56 at 16 weeks (explicit paper figure) added; 129×C57BL/6J F1 hybrid also unaffected.
Horai 2000 cytokines corrected: body now specifies IL-1β, IL-6, and TNF-α mRNA elevation in joints (matches paper’s Northern blot data); prior text said “elevated TNF and IL-1β” (omitting IL-6).
Van Tassell 2014 footnote: added “14-day treatment periods” crossover design detail; CRP p=0.006 flagged as not independently confirmed from full text.
Limitations section updated to reflect actual archive status: Larsen + Horai now verified; Van Tassell partial; closed-access four sources listed explicitly.

Unverifiable claims (closed-access sources):

Eisenberg 1990 / Hannum 1990: cloning details, agonist-activity characterisation
Arend 1998: isoform descriptions (4-isoform table), molar-ratio requirement (100–1000x), in-vivo roles
Bresnihan 1998: RA phase 3 ACR20 43% vs 27%, radiographic progression slowing, FDA pivotal data

Van Tassell CRP p-value (p=0.006): consistent with abstract summary but not confirmed against primary PDF.

Downstream pages to check (main agent):

None identified — il-1ra.md is an entity page, not a study page; corrections here originate at this page and do not propagate upstream. The anakinra compound page (stub) will eventually inherit these trial descriptions; verify when seeded.

[2026-05-06] R25 verifier sweep (18 pages)

All 18 R25-seeded pages passed through the wiki-verifier. Every page now ships verified: true with a verified-scope documenting which sources were PDF-verified end-to-end vs left at no-fulltext-access. Most pages required substantive corrections — many catching either fabricated content or canonical-DB drift.

Major scientific corrections caught:

fibroadipogenic-progenitors: Entire fabricated “Senescent FAPs in chronic muscle disease” section removed. Reggio 2020 does NOT describe p16+/p21+/SA-β-gal+ FAPs, NO SASP profile, NO ABT-263/navitoclax test — these were AI-extraction fabrications. Paper is actually about WNT5a/GSK3/β-catenin axis. Wosczyna 2019 DOI corrected (07.073 → 04.074); ablation system corrected (DTR → PDGFRα^CreER;R26^DTA); satellite-cell timeline corrected (“weeks” → “~45% reduction over 9 months”). Heredia 2013 mouse model corrected (IL-5-null → ΔdblGATA mice); rescue experiment corrected (IL-4 injection → IL-4 complex). Joe 2010 isolation gate corrected (PDGFRα was validation, not primary). Uezumi 2010 mouse exclusion marker corrected (CD56 → SM/C-2.6).
elamipretide: Major TAZPOWER trial corrections: duration 4 weeks → 12 weeks; primary endpoint description (single-composite → two co-primary 6MWT + BTHS-SA); double-blind secondary endpoints all NS (originally claimed improvement); author list corrected (Thompson WR, not “Collier JJ et al.”); mean age 17 → 19.5 yr. PubChem CID confirmation: 11764719 / C32H49N9O5 / 639.8 Da. Mitchell 2022 mechanism re-characterized (membrane electrostatic potential, not “ROS-suppression potency”). Dmt electron-donating attribution moved Szeto 2014 (not Mitchell 2022).
heart: LV wall thickness “~25-30%” claim REMOVED — Strait & Lakatta 2012 explicitly states LV mass does NOT increase with healthy aging (decreases in men, unchanged in women). SAN absolute cell count “10,000–20,000” REMOVED (not in source); replaced with verified <10% relative figure. Aortic sclerosis prevalence figures unsourced. Go 2001 AF age bracket corrected (20-39: 0.1% → <55: 0.1%).
gpr109a: Tunaru 2003 footnote — wrong receptor isoform attribution (HM74A → HM74; HM74A is the Wise 2003 distinction). EC50 values disambiguated by isoform/assay (~0.1 µM HM74A cAMP vs ~1 µM HM74 / ~3 µM mouse PUMA-G Ca2+). Singh 2014 quantitative Treg data added (25.8% WT vs 14.4% Niacr1-/-; all KO mice died by day 10 of 3% DSS).
il-1r1: Greenfeder 1995 misattribution — beta-bulge/QGEESN swap experiment is NOT in this paper; paper only shows IL-1Ra fails to co-IP with IL-1RAcP. Beta-bulge claim now tagged needs-citation pending crystallographic source. CANTOS trial design corrected from “150 mg q3m vs placebo” → 4-arm trial (50/150/300 mg + placebo; 150 mg only arm meeting prespecified threshold).
il-1a: Werman 2004 / Buryskova 2004 attribution swap — NLS-deletion mutant experiment was Buryskova 2004’s, not Werman 2004’s. Buryskova 2004 HAT activity claim corrected (only p300 not CBP/p300; no direct HAT measurement). Horai 1998 unsourced contact-hypersensitivity claim REMOVED. Orjalo 2009 quantitatives added (79% senescent surface labeling, antibody concentrations, C/EBPβ co-regulator). Senescence models expanded 2 → 4 (bleomycin primary).
il-1ra: Larsen 2007 p-value precision (P=0.03 with 95% CI). Horai 2000 arthritis onset corrected (“by 8 weeks” → “5 weeks first, ~80% by 8wk, 100% by 13wk”). “Disease requires T cells” REMOVED (no T-cell depletion experiments in paper). Cytokine list expanded (IL-6 added). Autoantibody types specified (anti-IgG/RF, anti-CII, anti-dsDNA).
il-18: Caspase-1 cleavage site corrected: Asp36/Ala37 → Asp35/Asn36 (both residue # AND P1’ identity wrong). Caspase-3 site: Asp71 → Asp69/Ile70. IL-18BP MW disambiguated (~20 kDa core, ~40 kDa glycosylated). IL-18BP Kd attribution moved Novick 1999 → Kaplanski 2018. Tadekinig n + SAEs added.
mff: AMPK phosphosites: Ser146 → Ser155 + Ser172 (Toyama 2016 identifies no Ser146 site). Phosphonull mutant SA2 = Ser155Ala/Ser172Ala. In-vivo validation source corrected (Drosophila NMJ → mouse cortical neurons via in utero electroporation). CCCP removed from AMPK-dependent fragmentation list. Mitophagy flux claim qualified (paper showed fission only). Gandre-Babbe HeLa not COS-7; overexpression NOT cause fragmentation. Loson 2013 redundancy mischaracterization corrected.
fis1: HGNC ID 18076 → 21689 (third canonical-DB drift in mito-dynamics cluster). Faitg 2019 muscle fiber type INVERTED (oxidative → glycolytic; FIS1 elevated in white gastrocnemius, not soleus). Qi 2012 experiment type misattributed (in-vitro only; no in-vivo). Ren 2017 tissue source corrected (AD patient skin fibroblasts, not brain).
mid49: CRITICAL TOPOLOGY CORRECTION: tail-anchored → N-terminally anchored (UniProt Q96C03 + Palmer 2011 Fig 1E-F: TM aa 23-43, N-terminus aa 1-22 in IMS). Open Targets API 5xx resolved → tier 3. Palmer 2013 cell type corrected (Mfn-DKO MEFs, not MFF/FIS1 DKO HEK293).
mid51: Same N-terminally-anchored topology fix (sister-page pattern). ADP-binding sites: 6 → 1 nucleotide-binding pocket per monomer. NTase fold homologue corrected (CTP-synthase → cGAS, RMSD 2.4 Å). MIDI compound target inverted (DRP1-C367, NOT MID49/MID51 surface). Loson 2013 triple-knockdown experiment FABRICATED — paper does not perform it. “OPA14” designation removed (not used in source). HGNC 25979 confirmed corrected.
oma1: OPA1 cleavage site inversion — wiki had OMA1=S2/YME1L=S1; correct is OMA1=S1/YME1L=S2 (confirmed against Ehses 2009 + Baker 2014 PDFs end-to-end; Ehses 2009 quote: “OMA1 as the peptidase responsible for the inducible cleavage of OPA1 at processing site S1”). HGNC 23858 → 29661 (4th drift in cluster). “Constitutively inactive” → “constitutively active at low basal levels” (Baker 2014). Active site topology: matrix → IMS-facing. Acin-Perez 2018 cardiac models corrected (isoproterenol, AngII/HFpEF, HFD — NOT pressure overload, IR, doxorubicin). Korwitz 2016 BUG-2 fully resolved: real DOI is 10.1083/jcb.201507022 (neurodegeneration paper, not cardiac). Author Quirós → Baker 2014.
yme1l: Independent confirmation of S2 (not S1) cleavage via Hartmann 2016 Fig 3C. HGNC 17175 → 12843 (5th drift). R206W → p.R149W (wrong residue identity in Mendelian disease section). Author Sprenger → Hartmann (B Hartmann is first author of the eLife paper). siRNA → stable shRNA (Stiburek 2012 methods). Complex I/IV impairment → Complex I -60% / Complex IV unaffected. PRELID1 source resolved (Potting 2013, doi:10.1016/j.cmet.2013.07.008).

Canonical-DB drift summary (cluster-wide pattern): Five HGNC IDs in the mito-dynamics cluster were wrong as originally seeded. UniProt REST cross-check at verification time recovered correct values:

fis1: HGNC 18076 → 21689
mid49 (MIEF2): HGNC 17920 → confirmed corrected at seed time
mid51 (MIEF1): HGNC 25979 → confirmed corrected at seed time
oma1: HGNC 23858 → 29661
yme1l: HGNC 17175 → 12843

Cross-cutting propagation done by main agent post-verification:

opa1 — flipped S1/S2 cleavage assignments throughout (table, body, Key interactors, footnote) per oma1+yme1l verifier convergent finding.
mitochondrial-dynamics pathway — updated MFF AMPK phosphosite reference (Ser146 → Ser155/Ser172).

Pages that retained verified: true with partial-scope (closed-access PDFs not readable):

heat-shock-response: Anckar 2011 PDF locally available but unread; Locke 1996, Heydari 1993, Hsu 2003, Akerfelt 2010 all not_oa.
elamipretide: Marcinek 2019 + Sabbah 2016 DOI lookup failed; López-Otín 2023 closed.
heart: Strait & Lakatta 2012 verified via PMC; Bergmann 2009 DOI lookup failed; Go 2001 + Lakatta & Levy 2003 closed-access.
fibroadipogenic-progenitors: Lemos 2015 not_oa.
All mito-dynamics pages: Otera 2010, Anand 2014, Wai 2015 closed-access.

Verification statistics:

18 pages: 18 verified
Quantitative claims revised at the source-text level: ~80+
Fabricated experiments / unsupported attributions caught and removed: ~12
Canonical-DB drifts caught: 5 HGNC + 2 NCBI-Gene + 2 Ensembl
DOI corrections caught at verifier time (post-seeder): Korwitz 2016 (neurodegeneration paper), Wosczyna 2019 (07.073 → 04.074)

Schema escalation surfaced:

The cell-ontology-id, gtex-aging-correlation, mr-causal-evidence fields were left null on most R25 pages with appropriate gap tags. The next lint pass should sweep for under-populated optional fields.
The druggability-tier was assigned by judgment on most pages due to Open Targets API 5xx during seeding; mid49 + mid51 verifiers confirmed Open Targets API now responsive — a sweep to populate druggability-tier on remaining mito-dynamics pages would close this gap.

🧬 Aging Wiki

Explorer

R25

log/R25.md — Round 25 entries

[2026-05-07] verify | interventions/pharmacological/telomerase-activators.md

[2026-05-07] verify | molecules/proteins/tpp1.md

[2026-05-07] verify | molecules/proteins/dkc1.md

[2026-05-07] verify | molecules/proteins/tin2.md

[2026-05-07] verify | molecules/proteins/trf2.md

[2026-05-07] verify | molecules/proteins/pot1.md

[2026-05-07] verify | processes/replicative-senescence.md

[2026-05-07] verify | molecules/proteins/terc.md

[2026-05-07] verify | tak1

[2026-05-07] verify | myd88

[2026-05-07] verify | traf6

[2026-05-07] verify | hsp90

[2026-05-07] verify | il-1-signaling + nf-kb cross-fix

[2026-05-07] verify | anakinra

[2026-05-07] verify | unfolded-protein-response

[2026-05-07] verify | waziry-2023 study

[2026-05-06] verify | interventions/lifestyle/methionine-restriction.md

[2026-05-06] verify | interventions/lifestyle/time-restricted-eating.md

[2026-05-06] verify | interventions/lifestyle/heat-exposure.md

[2026-05-06] verify | interventions/pharmacological/nad-precursors.md

[2026-05-06] verify | interventions/pharmacological/ampk-activators.md

[2026-05-06] verify | interventions/gene-therapy/crispr-base-editing-pcsk9.md

[2026-05-06] verify | interventions/gene-therapy/aav-klotho.md

[2026-05-06] verify | interventions/gene-therapy/aav-follistatin.md

[2026-05-06] verify | interventions/gene-therapy/aav-osk.md

[2026-05-06] verify | interventions/stem-cell-therapy/hematopoietic-stem-cell-transplantation.md

[2026-05-06] verify | interventions/stem-cell-therapy/in-vivo-partial-reprogramming-therapy.md

[2026-05-06] verify | interventions/stem-cell-therapy/ipsc-derived-cell-therapy.md

[2026-05-06] verify | molecules/proteins/caspase-1.md

[2026-05-06] verify | cell-types/astrocytes.md

[2026-05-06] verify | molecules/proteins/gsdmd.md

[2026-05-06] verify | molecules/proteins/asc.md

[2026-05-06] verify | processes/pyroptosis.md

[2026-05-06] verify | cell-types/neurons.md

[2026-05-06] verify | molecules/proteins/asxl1.md

[2026-05-06] verify | molecules/proteins/dnmt3a.md

[2026-05-06] verify | hypotheses/mitohormesis.md

[2026-05-06] verify | molecules/proteins/fgf21.md

[2026-05-06] verify | molecules/proteins/myostatin.md

[2026-05-06] verify | molecules/proteins/hsf1.md

[2026-05-06] verify — molecules/compounds/coq10.md

[2026-05-06] verify | molecules/proteins/pcsk9.md

[2026-05-06] verify — molecules/compounds/paroxetine.md

[2026-05-06] verify — molecules/proteins/asm.md

[2026-05-06] verify — molecules/proteins/mid51.md (R25 Tier B)

[2026-05-06] verify — molecules/proteins/fis1.md (R25 Tier B)

[2026-05-06] verify — molecules/proteins/mid49.md (R25 Tier B)

[2026-05-06] verify — molecules/proteins/mff.md (R25 Tier B)

[2026-05-06] verify — molecules/proteins/yme1l.md (R25 Tier B)

[2026-05-06] verify — molecules/proteins/il-18.md (R25 Tier B)

[2026-05-06] verify — molecules/proteins/il-1a.md (R25 Tier B; partial PDF pass)

[2026-05-06] verify — molecules/proteins/il-1r1.md (R25 Tier B; partial PDF pass)

[2026-05-06] verify — cell-types/fibroadipogenic-progenitors.md (full PDF pass)

[2026-05-06] verify — tissues/heart.md (full PDF re-pass)

[2026-05-06] verify — cell-types/fibroadipogenic-progenitors.md

[2026-05-06] verify — molecules/proteins/oma1.md (R25 Tier B)

[2026-05-06] R25 — Atomic-content backlog (Tier A + Tier B; 18 pages)

[2026-05-06] verify — molecules/proteins/gpr109a.md

[2026-05-06] verify — molecules/proteins/il-1ra.md (R25 Tier B; partial PDF pass)

[2026-05-06] R25 verifier sweep (18 pages)

Graph View

Table of Contents

Backlinks