R24

log/R24.md — Round 24 entries

Sub-file of log — see parent for index.

[2026-05-06] R24 — Atomic-backlog completion + R23 follow-on + schema cleanup

Round complete. 17 atomic pages + 5 CLAUDE.md schema edits.

Pages added:

• R24a (4 proteins): molecules/proteins/{myostatin, pcsk9, hsf1, fgf21}.md
• R24b (4 CHIP-triad proteins): molecules/proteins/{dnmt3a, asxl1, tet1, tet3}.md
• R24c (3 cell types): cell-types/{neurons, astrocytes, oligodendrocytes}.md
• R24d (3 proteins + 1 process): molecules/proteins/{caspase-1, asc, gsdmd}.md + processes/pyroptosis.md
• R24e (2 hypotheses): hypotheses/{mitohormesis, antagonistic-pleiotropy}.md

Pages updated (propagation):

• phenotypes/clonal-hematopoiesis.md — “implicit stub — no page yet” notes for DNMT3A/ASXL1/TET2 cleaned up; gained TET1/TET3 family-completion links
• hallmarks/chronic-inflammation.md — NLRP3-inflammasome subsection enriched with caspase-1 / ASC / GSDMD / pyroptosis cross-links
• (R24a → R23 back-links already present from seeder wikilink-to-implicit-stub pattern; no main-agent edits needed)

CLAUDE.md edits (5 of 9 escalations addressed; 4 already done):

1. type: cell-type block — added verified / verified-date / verified-by / verified-scope fields
2. type: cell-type block — added body note for iPSC schema exception (empty tissue-of-origin: + key-aging-phenotypes:)
3. safety-profile field — annotated indication-split convention (pipe-delimited free-text)
4. human-evidence-level enum — added limited-negative value (STAC-class precedent)
5. clinical-stage field — annotated class-page convention (aging-context vs. maximum-regulatory)

(Items 6-9 — type:framework verified strip, type: microbe formalization, biologic: true + who-inn:, retracted-paper convention — already in CLAUDE.md from prior rounds.)

DOI corrections caught at seed time (5):

• Hoischen 2011 ASXL1: .870 → .868
• Inagaki 2007 FGF21: .05.002 → .05.003 (.05.002 is Badman 2007)
• Beck 2020 TET3: .2020.06.020 → .2019.12.007
• “Bhat 2012 Cell” → Bhat 2012 PLoS ONE (different paper)
• Two TET1 DOIs caught wrong in brief, corrected via PubMed/Crossref

Critical scientific correction: TET3 seeder caught my brief’s incorrect statement that TET3 lacks the CXXC domain. UniProt confirms TET3 has CXXC; TET2 is the family member without CXXC. Documented in TET3 page body.

Schema escalations surfaced for R25+:

• DNMT3A seeder: mr-causal-evidence field semantics need extension for somatic-mutation proteins (CHIP variants are somatic, not germline → standard MR doesn’t apply). Consider mr-not-applicable-somatic enum value.
• Antagonistic-pleiotropy seeder: key-evidence-for / key-evidence-against may not apply to Mode B conceptual-frame hypotheses (SOP omits these from Mode B template but schema lists them). Lint pass should consider Mode-B-specific suppression.
• Multiple R24 seeders flagged closed-access primary sources (Williams 1957, Hamilton 1966, Kirkwood 1977, Mannick 2014/2018, Cohen 2006); verifier sweep will need access acquisition before flipping verified: true.

Verification status: All 17 pages ship verified: false with ⚠️ banner; await wiki-verifier sweep. DOI sanity at seed time confirmed.

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

Sources checked:

• doi:10.1038/415045a (Tyner SD et al. 2002, Nature) — PDF local; verified end-to-end (4 pages)
• doi:10.1093/emph/eoy033 (Austad SN & Hoffman JM 2018, Evol Med Public Health) — PDF downloaded during session (was status: pending; OA gold PMC); verified end-to-end (8 pages)
• doi:10.1111/j.1558-5646.1957.tb02911.x (Williams 1957) — not_oa; closed-access; unverifiable
• doi:10.1016/0022-5193(66)90184-6 (Hamilton 1966) — not_oa; closed-access; unverifiable
• doi:10.1038/270301a0 (Kirkwood 1977) — not_oa; closed-access; unverifiable

Cross-link resolution:

• p53 — verified:true
• insulin-igf1 — verified:true
• mtor — verified:true
• p53-pathway — verified:true
• disposable-soma-theory — verified:true
• free-radical-theory-of-aging — verified:true
• information-theory-of-aging — verified:true
• telomere-attrition — verified:true
• cellular-senescence — verified:true (link corrected from cellular-senescence which was a missing page)

Corrections made:

1. Tyner 2002 allele names — CRITICAL CORRECTION. Wiki described the experiment as “the ‘super-p53’ and ‘Δ40p53’ mouse experiments.” Neither term appears in Tyner 2002. The paper describes p53+/m mice (m-allele: deletion of exons 1–6, expressing a C-terminal p53 fragment that confers constitutive activation-like phenotypes consistent with activated p53) and a second line carrying a temperature-sensitive mutant allele (pL53). “Super-p53” is García-Cao et al. 2002 (a different paper, extra copies of the WT locus); “Δ40p53” designates an N-terminal truncation isoform from different literature. Body text corrected to describe the p53+/m m-allele and pL53 allele accurately, with confirmed numerics (median lifespan 96 wk vs 118 wk p53+/+; tumour incidence <6% vs >45%). Footnote updated accordingly. [Scientifically critical — wrong allele nomenclature attributed to paper.]
2. Fabricated Austad & Hoffman 2018 direct quote — REMOVED. Wiki put in quotation marks: “the danger with AP is that it is used as an explanation for any gene that has deleterious effects in old age, regardless of whether early benefits have been demonstrated.” This exact phrase does not appear in the paper. The concern is made substantively in the paper, but without that phrasing. Replaced with accurate paraphrase of the paper’s actual discussion (many lifespan-extension lab studies do not report early fitness components; natural-population AP alleles remain largely undiscovered). [No fabricated quotes in verified pages.]
3. Austad & Hoffman 2018 overall conclusion characterization — CORRECTED. Wiki framed the paper primarily as a skeptical critique. The paper’s actual conclusion is more supportive: “whenever an allele or new mutation is discovered to extend life, some detrimental effect on early life fitness is almost always observed.” The paper concludes AP is “somewhere between very common or ubiquitous throughout the animal world.” Body text updated to reflect this nuanced-supportive conclusion alongside the caveats the paper raises. [Framing correction.]
4. Broken wikilink cellular-senescence → cellular-senescence. The processes/ path does not exist; actual page is at hallmarks/. Fixed in two body locations. [Cross-link integrity.]
5. Austad 2018 footnote gap tag removed. Was tagged no-fulltext-access (download pending at seeding time). PDF now downloaded and verified; tag removed and conclusion summary added to footnote.

Unverifiable claims (closed-access sources):

• Williams 1957 framing claims (AP as complement to Medawar mutation accumulation; declining force of selection mechanism) — based on widely-cited secondary sources. no-fulltext-access preserved.
• Hamilton 1966 “selection shadow” formalization — same. no-fulltext-access preserved.
• Kirkwood 1977 disposable soma / AP complementarity acknowledgment — same. no-fulltext-access preserved.
• APOE4 early-life benefit claims — not directly covered by any PDF read in this pass; needs-replication preserved.

Mode-B framing assessment: Appropriate. Page correctly uses treatment-mode: conceptual-frame and status: active-frame. Structure follows Mode B SOP template. key-evidence-for: array points to verified atomic pages (p53, insulin-igf1, mtor) — permissible per R26a convention for Mode B (not required but not prohibited). No primary quantitative claims originate on this page except Tyner 2002 numerics (now corrected and footnote-sourced). Austad 2018 discussion is framing/synthesis, not primary claims.

Downstream propagation needed:

• p53-pathway — may describe Tyner 2002 using “super-p53” or “Δ40p53” allele terminology; check and propagate corrected allele names (p53+/m m-allele, pL53 temperature-sensitive allele).
• cellular-senescence — AP trade-off framing for senescence as tumor-suppressor vs SASP driver is now confirmed consistent; no numeric corrections needed, but the cross-link from this page should be validated in next lint pass.
• disposable-soma-theory — cites AP/DS complementarity via Kirkwood 1977; that claim is unverified (closed-access) but is a standard historical characterization; no action required from this pass.

Verified state: verified: true with verified-scope documenting Tyner 2002 + Austad 2018 PDF verification, Williams/Hamilton/Kirkwood remaining closed-access, and cross-link integrity check.

[2026-05-06] verify | molecules/proteins/tet1.md

Sources checked:

• doi:10.1126/science.1170116 (Tahiliani et al., Science 2009) — download failed (status: failed after retry); verified via PMC full-text (PMC2715015); green OA
• doi:10.1038/nature09303 (Ito et al., Nature 2010) — PDF verified end-to-end (local archive)
• doi:10.1016/j.stem.2011.01.008 (Koh et al., Cell Stem Cell 2011) — PDF verified end-to-end (downloaded on demand)
• doi:10.1016/j.neuron.2013.08.003 (Rudenko et al., Neuron 2013) — PDF verified end-to-end (downloaded on demand)
• doi:10.1002/anie.201502722 (Wagner et al., Angew Chem Int Ed 2015) — PDF verified end-to-end (downloaded on demand)
• doi:10.1016/j.ccr.2010.11.015 (Figueroa et al., Cancer Cell 2010) — PDF verified end-to-end (downloaded on demand)
• doi:10.1038/nature12362 (Blaschke et al., Nature 2013) — not_oa; claims confirmed via Crossref metadata and cross-reference in minor2013 text; no-fulltext-access for blaschke2013 specifically
• doi:10.1074/jbc.C113.464800 (Minor et al., J Biol Chem 2013) — PDF verified end-to-end (downloaded on demand)

Corrections made:

• wagner2015 aging-direction (SCIENTIFICALLY CRITICAL): “age-dependent reductions in 5hmC” → corrected to: Wagner 2015 documents 5hmC increase during postnatal brain development, reaching steady state ~1.2% in adult human grey matter by ~age 22; adult aging decline not demonstrated (n=2–4 for oldest subjects). Aging-decline framing removed from body text and footnote. Extrapolation table updated to reflect ambiguity.
• rudenko2013 behavioral paradigm: “memory extinction deficits in Morris water maze and fear-conditioning” → clarified that memory acquisition is intact; extinction deficits demonstrated in both contextual fear conditioning (n=10+9; p<0.01) and spatial MWM (n=10+9); added Npas4 promoter hypermethylation quantitative data (~20–45% vs ~3.5–8% in controls); added LTD finding; removed misleading use of “learning” (acquisition is intact)
• figueroa2010 scope: “potently inhibits TET1/2” → corrected to: Figueroa 2010 demonstrates inhibition of TET2 directly; TET1 inhibition by 2-HG is mechanistically plausible but not directly demonstrated in this paper
• minor2013 footnote model: “HEK293 + cell-free system” → corrected to: primary model MEFs (C57BL/6), with HEK-293T and HeLa as confirmatory; no cell-free system used; added 4-fold increase quantitative result
• rudenko2013 footnote: added n per group, strain, age, p-value, methylation percentages, LTD finding
• wagner2015 footnote: replaced vague “age-dependent decline” with accurate description of developmental increase and sample sizes
• Limitations section: updated to accurately describe Wagner 2015 as showing developmental accumulation, not aging decline
• Flipped verified: true; removed auto-extraction banner

Unverifiable claims:

• blaschke2013 (doi:10.1038/nature12362): not_oa; wiki claims “vitamin C drives Tet-dependent global demethylation; effect abolished in Tet1/Tet2 double KO” — mechanistically consistent with minor2013 and referenced in minor2013 text, but full PDF not verified. Tagged implicitly by verified-scope note.
• Citation counts in footnotes (~5,700×, ~2,500×, etc.) not independently re-verified.
• UniProt identity fields (Q8NFU7, length 2,136 aa, MW ~235 kDa) not re-checked against UniProt live database.

Downstream propagation needed (main agent):

• molecules/proteins/tet2.md — if it inherits the “5hmC decline with aging” framing from Wagner 2015, that claim needs the same correction
• molecules/proteins/tet3.md — R24b sibling stub; if seeded with similar aging narrative, same correction applies
• hallmarks/epigenetic-alterations.md — likely cites 5hmC brain decline as an aging biomarker; verify whether it cites Wagner 2015 or a more appropriate source for the aging-decline claim
• processes/clonal-hematopoiesis.md or [[clonal-hematopoiesis]] — figueroa2010 TET2/IDH2 context; if it characterizes 2-HG as inhibiting “TET1/2,” narrow to TET2 (directly demonstrated) with TET1 as plausible-but-undemonstrated

[2026-05-06] verify | molecules/proteins/tet3.md

Sources checked:

• doi:10.1016/j.ajhg.2019.12.007 (Beck 2020, Am J Hum Genet) — PDF downloaded and verified end-to-end (bronze OA; 12 pages)
• doi:10.1101/gad.276568.115 (Rasmussen & Helin 2016, Genes Dev) — PDF downloaded and verified end-to-end (diamond OA; 18 pages)
• doi:10.1073/pnas.1318906111 (Li 2014, PNAS) — PDF downloaded and verified end-to-end (bronze OA; 6 pages)
• doi:10.1016/j.neurobiolaging.2012.02.006 (Dzitoyeva 2012, Neurobiol Aging) — green OA via PMC3462297; PDF download failed (0 candidate URLs); verified via PMC full-text HTML
• doi:10.1038/nature10443 (Gu 2011, Nature) — not_oa; PubMed abstract only; zygotic reprogramming claims unverifiable from full PDF
• UniProt O43151 — canonical length (1,795 aa), MW (~193.7 kDa), CXXC domain (residues 50-90) verified via REST API

Known-at-seed corrections confirmed in place:

• Beck 2020 DOI is 10.1016/j.ajhg.2019.12.007 (not.2020.06.020) — confirmed correct in frontmatter and footnote
• TET3 retains CXXC (residues 50-90); TET2 lacks it — confirmed correct; Rasmussen 2016 Fig. 1 + text confirm
• HGNC R26d correction (11591 to 28313) — confirmed in frontmatter and Identity section

Corrections made:

1. Dzitoyeva 2012 n: n~6 -> n=5 per age group. PMC full-text: “5 adult (4 months) and 5 old (24 months) C57BL/6 male mice.” [Numeric correction.]
2. Dzitoyeva 2012 ages: “3 vs 24 months” -> “4 months vs 24 months.” Adult group is 4 months, not 3 months. [Factual correction.]
3. Dzitoyeva 2012: “TET1 and TET3 protein levels altered” -> TET2 and TET3 mRNA increased in cerebellum. Paper measured mRNA not protein; frontal cortex showed no significant TET mRNA changes with aging; cerebellum showed TET2+TET3 mRNA increase. [Scientifically critical — protein vs. mRNA conflation, wrong brain region, wrong directionality.]
4. Dzitoyeva 2012 body rewritten. Removed “links age-related 5hmC decline to TET-family downregulation.” Frontal cortex 5hmC declines with no TET mRNA change; cerebellum 5hmC increases with TET mRNA increase. Causal mechanism unresolved. [Scientifically critical — mechanism overstatement removed.]
5. Li 2014 n: n10 -> n=3-8 per experiment. n=3/group in vitro; n=4-5/group in vivo molecular; n=7-8/group behavioral. “n10” matches no reported group size. [Numeric correction.]
6. Li 2014 body: “TET3 protein levels rise” -> “Tet3 mRNA rises.” Li 2014 measures Tet3 mRNA as activity-dependent signal; protein not directly assayed in key results. [Protein vs. mRNA conflation.]
7. Li 2014 body: gephyrin intron detail added. Paper demonstrates TET3 occupancy and 5hmC accumulation within the gephyrin gene intron (Fig. 3F).
8. Beck 2020 body: “in the catalytic domain” -> “in or near the catalytic domain.” p.Arg752Cys (family 1) is upstream of the catalytic domain (aa 773-1776) per Fig. 2A. [Factual overgeneralization.]
9. Beck 2020 footnote: assay corrected from “fibroblasts” to “HEK293 cells.” Functional 5hmC assay used HEK293 overexpression. Added 8-family count; added autistic features and movement disorders to phenotype list. [Factual correction to study design.]
10. Tet1/Tet2 double KO table: “Viable” -> “Mostly perinatal lethal.” Rasmussen 2016 (citing Dawlaty 2013): “the majority die perinatally” with encephaly, growth retardation, and compromised imprinting. [Scientifically critical — wrong viability claim corrected.]
11. Rasmussen 2016 and Li 2014 footnotes updated from “PDF pending” to “PDF downloaded 2026-05-06”; Rasmussen footnote enriched with key findings confirmed in PDF.
12. Auto-extraction banner removed; verified flag flipped to verified: true.

Unverifiable claims:

• Gu 2011 (not_oa): n~60 embryos, STELLA/PGC7 protection, implantation failure rates — retained with archive-status note; full PDF not accessible
• Rasmussen 2016 Fig. 1 shows “TET3 (1660 aa)” vs UniProt canonical 1,795 aa — discrepancy noted; UniProt canonical used as authoritative; no correction made

Downstream propagation needed (main agent):

• molecules/proteins/tet2.md — confirm TET2 CXXC loss stated correctly; confirm page does not use “TET protein downregulation drives 5hmC aging decline” framing that Dzitoyeva 2012 correction now contradicts
• hallmarks/epigenetic-alterations.md — check for “TET protein downregulation drives neuronal 5hmC loss with aging” language now invalidated by Dzitoyeva 2012 correction

Verified state: verified: true with verified-scope documenting PDF vs PMC-fulltext vs abstract-only tiers. R26d HGNC (11591 to 28313) and CXXC corrections confirmed in place.

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

Sources checked:

• doi:10.1523/jneurosci.22-07-02451.2002 (Sim et al., J Neurosci 2002) — PDF verified end-to-end (downloaded on demand; PMC OA)
• doi:10.1073/pnas.1818348116 (Nicaise et al., PNAS 2019) — PDF verified end-to-end (downloaded on demand; PMC OA)
• doi:10.1016/j.stem.2019.08.015 (Neumann et al., Cell Stem Cell 2019) — PDF verified end-to-end (downloaded on demand; OA)
• doi:10.1038/nrn.2017.136 (Franklin & ffrench-Constant, Nat Rev Neurosci 2017) — PDF download failed (status: failed, OA=green); abstract confirmed via Europe PMC; body claims not individually verifiable
• doi:10.1016/j.neurobiolaging.2009.08.007 (Bartzokis, Neurobiol Aging 2011) — PDF download failed (status: failed, OA=green); abstract confirmed via Europe PMC; ~10% white-matter volume figure not in abstract
• doi:10.1016/s0140-6736(17)32346-2 (Green et al., Lancet 2017, ReBUILD trial) — PDF download failed (status: failed, OA=bronze); verified via PubMed abstract (PMID 29029896)

Corrections made:

1. (SCIENTIFICALLY CRITICAL) Green 2017 p-value: p=0.0288 → p=0.0048 (95% CI 0.5–2.9 ms/eye). Corrected in body text and footnote.
2. (SCIENTIFICALLY CRITICAL) Sim 2002 young animal age: “3-month” → “8–10-week-old.” Methods state young = female Sprague-Dawley rats 8–10 weeks; aged = ≥12 months.
3. Sim 2002 demyelination model: “lysolecithin” → “ethidium bromide” into caudal cerebellar peduncle.
4. Sim 2002 mechanism reframed: “differentiation block is the rate-limiting step” contradicted the paper. Rewritten to reflect BOTH recruitment impairment (lower OPC density at 5/7/10 dpl; p<0.05) AND widening differentiation latency; OPC numbers in intact tissue not different between age groups.
5. Neumann 2019 OPC isolation ages: “aged (12-month)” → “≥18-month-old (typically 20–24-month-old)” for in vitro OPC work.
6. Neumann 2019 mechanism: “fatty-acid oxidation” → “restored mitochondrial oxidative capacity (OCR and ATP levels)” via AMPK.
7. Neumann 2019 primary intervention: ADF is the primary in vivo intervention; metformin is the fasting mimetic. Section rewritten.
8. Nicaise 2019 cell-type precision: “senescent NPCs/OPCs” → “senescent SOX2+ NPCs”; OPCs are downstream targets. Footnote updated (iPSC-derived NPCs from PPMS patients, not rat OPC culture).
9. Green 2017 population and dose added: relapsing MS with chronic demyelinating optic neuropathy; clemastine fumarate 5.36 mg twice daily; 150-day crossover; 95% CI added to footnote.
10. Removed auto-extraction banner; flipped verified: true with verified-scope.

Unverifiable claims (PDF-inaccessible sources):

• ~10% white-matter volume loss by the eighth decade (bartzokis2011 body claim) — not in abstract; retained with needs-replication.
• Franklin 2017 body claims (MCT1/MCT2 lactate delivery, BDNF/GDNF/IGF-1 from OLs, ER stress in aged OLs) — abstract confirms review scope; individual body statements not verifiable from abstract alone.

Downstream propagation needed (main agent):

• molecules/compounds/metformin.md — check Neumann 2019 mechanism; “fatty-acid oxidation” is incorrect (correct: AMPK → mitochondrial OCR/ATP restoration).
• interventions/pharmacological/senolytics.md — if Nicaise 2019 cited, check cell-type description (“OPCs” imprecise; should be “SOX2+ NPCs”).
• hallmarks/cellular-senescence.md — may cite Nicaise 2019; propagate NPC/OPC cell-type precision if needed.
• hallmarks/stem-cell-exhaustion.md — may cite Sim 2002; check for “3-month” young age or “lysolecithin” model.

[2026-05-06] R26b — R24 atomic-page verification sweep (DONE; 17/17 pages)

All 17 R24 atomic pages verified. Each page now ships verified: true with verified-scope line documenting PDF-verified vs partial/abstract-only sources. Auto-extraction banners removed.

Pages verified (sub-batch / page → critical-correction one-liner):

R24a — R23 follow-on protein atoms:

• myostatin — Schuelke 2004 mutation type changed (frameshift → splice-site IVS1+5 g→a + NMD); Rooks 2017 misattributed as “sarcopenia 6MWT trial” (actually disuse-atrophy proof-of-concept in healthy young men, n=24); SMAD2/3 → atrogin-1/MuRF1 over-attribution corrected; bimagrumab dosing confirmed (single 30 mg/kg i.v., +5.1% TMV vs +0.1% placebo at 12 wk).
• pcsk9 — autocatalytic cleavage Gln152↓Asn153 → Gln152↓Ser153 (VFAQ:SIP); EGF-A binding domain corrected (CRD → catalytic domain 367-381); Domain table P-domain/CHRD split; ODYSSEY MACE event rates inverted (1.7% alirocumab vs 3.3% placebo); Y142X MAF 2.6% → 0.8% (combined carrier 2.6%); secreted form 62 → 65 kDa.
• hsf1 — basal localization changed cytoplasm/nucleus → predominantly nuclear (Anckar 2011 p.1091 explicit); HSF1A citation stripped from Anckar (correct source = Neef 2010); domain residues tightened (DBD 14-122→15-110; HR-A/B 130-230→130-221; TAD 420-529→410-529); Morley 2004 lifespan precision (22% extension, 16.8 vs 13.8 d, p<0.001).
• fgf21 — Beta-Klotho co-receptor wrongly attributed to Kharitonenkov 2005 (paper says “currently unclear”); PI3K-AKT signaling claim removed (paper shows MAPK/ERK only); PPARα-null fails-to-induce-FGF21 overstatement corrected (~5-fold null vs ~28-fold WT, n=4); Zhang 2012 Tg cohort exact n added (Tg=77, WT=67); Laeger 2014 conflated with methionine-restriction (paper uses general low-protein diet ~5% casein).

R24b — CHIP triad completion:

• dnmt3a — Genovese 2014 CHIP fraction table (~22%/~17%/~10%) not in source; replaced with raw counts (190 / 35 / 31 of 327 driver mutations) + Jaiswal raw counts; Jaiswal CHD CI corrected; Yan 2011 scope corrected (AML-M5 only 20.5%, not pan-AML “20-26%”); R882H mechanism (~88% activity reduction; co-expression required, NOT in-vitro mixing); mr-causal-evidence: not-applicable-somatic per R26a.
• asxl1 — Genovese hematologic-cancer HR (11x → 12.9 [5.8-28.7]; “11x” belonged to Jaiswal); Abdel-Wahab 2013 critical: Asxl1+Tet2 double-KO does NOT accelerate AML (paper shows opposite — Tet2 rescues self-renewal defect, MDS-like not AML); Gumuser 2023 cohort (UK Biobank → ASCVD); mr-causal-evidence: not-applicable-somatic.
• tet1 — Wagner 2015 aging-direction misattribution corrected (paper shows postnatal increase to ~1.2% in adult brain by ~age 22; aging decline NOT demonstrated, n=2-4 oldest subjects); Rudenko 2013 behavioral framing corrected (extinction deficits, NOT learning impairment); Figueroa 2010 scope narrowed (TET2 only, not TET1/2); Minor 2013 model corrected (MEFs primary, not HEK293).
• tet3 — Dzitoyeva 2012 protein-vs-mRNA conflation throughout (mRNA only); Dzitoyeva n=5/group (not ≈6), ages 4 vs 24 mo (not 3 vs 24); Beck 2020 catalytic-domain qualifier (in or near, p.Arg752Cys upstream); Tet1/Tet2 double-KO viability (“viable” → “mostly perinatal lethal” per Rasmussen 2016/Dawlaty 2013); R26d HGNC fix (11591→28313) confirmed; CXXC retention fix (TET3 retains, TET2 lacks) confirmed.

R24c — Neuro cell-types:

• neurons — Jurk 2012 organism rat → mouse C57BL/6 (paper used mice exclusively); Lodato 2018 cohort breakdown (15 normal donors, 93 PFC + 26 DG + 42 disease neurons; 161 total); Lodato disease attribution AD → Cockayne syndrome + Xeroderma pigmentosum (paper has no AD data); SNV burden replaced (~2000 → fold-excess ~2.3× / ~2.5×); ~40 SNVs/neuron/year DG rate added.
• astrocytes — 60% interstitial-space-during-sleep figure misattributed to Iliff 2012; correct source is Xie 2013 (sleep glymphatic); Clarke 2018 age framing (12 mo → 2 yr) and A1 marker list (C3/C1s/SERPING1 → Serpina3n/C4B/C3/Cxcl10 — the validated four); Bhat 2012 p21 in-vivo claim removed (paper uses p16 only); A1/A2 binary deprecated per Escartin 2021 consensus.
• oligodendrocytes — Sim 2002 demyelination model lysolecithin → ethidium bromide; Sim 2002 mechanism corrected (both recruitment and differentiation impaired, not just differentiation); Neumann 2019 OPC isolation ages (12 mo → ≥18 mo); Neumann mechanism (AMPK→FAO → AMPK→OCR/ATP); Nicaise 2019 cell type (NPC/OPC → SOX2+ NPCs; OPCs are downstream); Green 2017 p-value 0.0288 → 0.0048.

R24d — Inflammasome closure:

• caspase-1 — IL-1β cleavage site Asp116-Ala117 → Asp116-Ser117 (UniProt P01584 confirmed); BHB mechanism (NLRP3 ATPase blocking → K⁺ efflux + ASC oligomerization inhibition; ATPase mechanism is MCC950’s, not BHB’s); Youm 2013 KO arm corrected (Casp1-/- not in paper; Nlrp3-/- and Asc-/- only); Camell 2017 GDF3 intermediate added; R26d Ensembl correction (TP53→CASP1) confirmed.
• asc — domain residues PYD 1-91→1-105, CARD 107-195→105-195 (Srinivasula 2002 explicit); Lu 2014 structural descriptor “left-handed helical” → “three-start C3 symmetry” (left-handed not in paper); Franklin 2014 propagation re-described (phagocytosis→lysosomal damage→de-novo nucleation; recipient response partially NLRP3-independent); Ganapathy 2025 model corrected (FMO3 KO primary, adipocyte-derived).
• gsdmd — Pore stoichiometry 16-mer/10-14 nm → 27-mer/18 nm (Hu 2020 / Liu 2016 / Pan 2018 cryo-EM; Galluzzi 2018 16-mer was earlier review estimate); DSF cellular dose 4 µM → IC50 7.7 µM canonical / 10.3 µM non-canonical THP-1/iBMDM; Mejias 2018 GSDMD measured in mouse brain tissue, NOT human serum (the human-serum arm measured ASC and IL-18 only); NSA mechanism (upstream caspase inhibition, not direct Cys-191/192); Cys species specificity (Cys-191 = human; Cys-192 = mouse).
• pyroptosis — GSDMD pore inner diameter ~10-20 nm → ~10-14 nm with 16 protomers per Galluzzi 2018 (reconciled with gsdmd verifier’s 27-mer/18-nm via cross-page note pointing to more-recent primary literature); CANTOS dose-specificity tightened (150 mg + 300 mg both significant); NCCD definition quote restored.

R24e — Hypothesis pages:

• mitohormesis (Mode A) — Lonidamine attribution moved from Ristow 2011 review to primary Schmeisser experiment (#gap/unsourced for the primary).
• antagonistic-pleiotropy (Mode B) — Tyner 2002 allele names corrected (super-p53 / Δ40p53 not in Tyner 2002; correct: p53+/m m-allele exons 1-6 deletion + pL53 ts-allele); fabricated Austad 2018 quote removed; broken [[processes/cellular-senescence]] → [[hallmarks/cellular-senescence]] (×2).

R26b sweep statistics:

• 17 pages verified
• 80+ source-text-level corrections
• ~15 scientifically critical (residue identity, organism mis-attribution, fabricated experiments, inverted endpoints)
• 3 canonical-DB drifts caught at R26d preceded the verifiers (MFF Ensembl, caspase-1 Ensembl, TET3 HGNC)
• Cross-page literature discrepancy surfaced: GSDMD pore stoichiometry — Galluzzi 2018 16-mer/10-14 nm vs Hu/Pan/Liu 2016-2020 27-mer/18 nm — both documented, primary-literature value preferred.