🧬 Aging Wiki

R32

36 min read

log/R32.md — Round 32 entries

Sub-file of log — see parent for index.

[2026-05-07] verify | molecules/compounds/akkermansia-supplementation

Pages verified: 1

  • molecules/compounds/akkermansia-supplementation.md — 8 factual corrections; AI-extraction banner removed; verified: true with scope (Plovier 2017 not_oa — no-fulltext-access retained)

Sources read end-to-end (full PDF):

  • Depommier 2019 (Nat Med, doi:10.1038/s41591-019-0495-2) — locally available; 14 pp. + extended data
  • Kang 2024 (Nutrients, doi:10.3390/nu16234037) — downloaded gold OA; 10 pp.
  • Everard 2013 (PNAS, doi:10.1073/pnas.1219451110) — downloaded bronze OA; 6 pp.
  • Forslund 2015 (Nature, doi:10.1038/nature15766) — downloaded OA; 7 pp.

Sources unverifiable (closed-access):

  • Plovier 2017 (Nat Med, doi:10.1038/nm.4236) — not_oa; framing verified via abstract only; no-fulltext-access retained for Amuc_1100 mechanistic claims

Corrections made:

  1. Depommier 2019 exclusion count: “excluded: 7 due to antibiotic use, personal reasons, or protocol violation” → 8 total excluded (7 withdrew before completion + 1 additional excluded from analysis for protocol violation); allocation numbers per group added (13/13/14 allocated)
  2. Depommier 2019 body weight p-value: P=0.09 → P=0.091 (per abstract)
  3. Depommier 2019 fat mass p-value: P=0.09 → P=0.092 (per abstract)
  4. Depommier 2019 plasma LPS: “Significantly reduced / P<0.05” → “Significantly reduced vs baseline and vs placebo; p-value not stated in text (see Fig. 3d)”
  5. Kang 2024 myostatin claim: “follistatin/myostatin ratio supports muscle mass maintenance” → myostatin not significantly changed (P=0.3145); only follistatin increased (P=0.0063); ratio claim removed
  6. Kang 2024 muscle results: total extensor significant results (P=0.049 peak torque; P=0.0346 TQ/BW) added; grip strength null result noted
  7. Everard 2013 dose: 1.5×10^8 → 2×10^8 cfu/day (corrected in supplementation page body, footnote, and propagated to akkermansia-muciniphila.md footnote + body)
  8. Forslund 2015 AKK-metformin claim: “higher Akkermansia abundance” in metformin-treated T2D → qualified to “trend in Danish (MHD) cohort only; inconsistent across CHN and SWE subsets; paper states ‘to some extent’”
  9. Clinical trials active count: 9 → 14 (re-verified against ClinicalTrials.gov v2 API 2026-05-07); full 14-trial table added

Downstream pages to update (main agent):

  • akkermansia-muciniphila — Everard 2013 dose already propagated (corrected in this pass); Forslund 2015 metformin-AKK wording may need qualification there too (currently says “metformin enriches AKK” without the cross-cohort inconsistency caveat)
  • everard-2013-akkermansia-hfd-mice — if this study page exists, dose needs correction from 1.5×10^8 to 2×10

[2026-05-07] verify | interventions/dietary/postbiotics

Pages verified: 1

  • interventions/dietary/postbiotics.md — 6 factual corrections; AI-extraction banner removed; verified: true with scope

Sources read end-to-end (full PDF):

  • Salminen 2021 (Nat Rev Gastroenterol Hepatol, doi:10.1038/s41575-021-00440-6) — ISAPP postbiotic consensus; 19 pp.; downloaded via PMC OA during session
  • Depommier 2019 (Nat Med, doi:10.1038/s41591-019-0495-2) — pasteurized A. muciniphila RCT; 28 pp.; locally available
  • Kang 2024 (Nutrients, doi:10.3390/nu16234037) — pasteurized A. muciniphila HB05 sarcopenia RCT; 10 pp.; downloaded via gold OA during session
  • Arai 2018 (PLoS ONE, doi:10.1371/journal.pone.0199018) — heat-killed L. paracasei MCC1849 mouse immunology study; 15 pp.; downloaded via gold OA during session

Corrections made:

  1. Depommier 2019 mean age: “~47 yr” → “~51 yr across groups (placebo 49.5±9.7, pasteurized 52.8±7.2, live 52.9±8.6 yr)” per Extended Data Table 1
  2. Depommier 2019 insulin sensitivity measurement method: “hyperinsulinaemic clamp” → “HOMA” — paper used Homeostasis Model Assessment, not a clamp
  3. Depommier 2019 LPS table: “Reduced ~35%” → “Significantly reduced vs baseline and vs placebo (no specific percentage stated in source)” — ~35% figure not present anywhere in paper text or figures
  4. Mechanism section: “serum LPS levels decreased ~35% in supplemented vs baseline” → specific percentage removed; wording corrected to match source
  5. Kang 2024 mean age: “not reported in abstract” → “mean age 65.03±3.83 yr” per Table 2; entry criterion SARC-CalF <11 added
  6. Arai 2018 mouse strain: “C57BL/6 mice” → “BALB/c mice (7 wk old, male SPF)” — factual strain error; BALB/c is immunologically distinct (Th2-biased) from C57BL/6

Unverifiable claims:

  • Clinical-trials count (3) and NCT numbers not independently re-verified against ClinicalTrials.gov API for this pass

Downstream pages to check (may cite corrected claims):

[2026-05-07] verify | interventions/dietary/prebiotics

Pages verified: 1 (partial scope — Chenhuichen 2022 closed-access)

  • interventions/dietary/prebiotics.md — 9 factual corrections/additions; AI-extraction banner removed; verified: true with scope

Sources read end-to-end (full PDF):

  • Gibson 2017 (Nature Rev Gastroenterol Hepatol, doi:10.1038/nrgastro.2017.75) — ISAPP prebiotic consensus; downloaded during session (hybrid OA, CORA mirror)
  • Ni Lochlainn 2024 (Nature Communications, doi:10.1038/s41467-024-46116-y) — PROMOTe RCT; locally available
  • Yang 2024 (JCI, doi:10.1172/JCI176507) — frailty prebiotic RCT; downloaded during session (gold OA)
  • Buigues 2016 (Int J Mol Sci, doi:10.3390/ijms17060932) — nursing home prebiotic RCT; downloaded during session (gold OA)

Sources unverifiable (closed-access):

  • Chenhuichen 2022 (Exp Gerontol, doi:10.1016/j.exger.2022.111809) — not_oa; title/authors confirmed via Crossref; body + footnote tagged no-fulltext-access

Corrections made:

  1. PROMOTe: vague “prebiotic supplement × trial duration” → exact dose 7.5 g/day inulin+FOS × 12 weeks
  2. PROMOTe: omitted co-intervention added: all participants received BCAA protein supplement + resistance exercise (critical for interpreting muscle null result)
  3. PROMOTe: “cognition significantly improved vs placebo (p<0.05)” → exact: cognitive first-factor score β=−0.482, 95% CI −0.813–−0.141, p=0.014; PAL errors β=7.55, 95% CI 4.65–10.46, p=0.001
  4. PROMOTe: chair rise time (primary endpoint) exact stats added: β=0.579, 95% CI −1.080–2.239, p=0.494; grip strength p=0.512; SPPB p=0.551
  5. PROMOTe: PAL test identified as the specific cognitive measure with the significant signal (not just “cognition” generically)
  6. Yang 2024: dose “not specified in abstract” → verified as 15 g/day (inulin + oligofructose 50:50, chicory-derived, after breakfast)
  7. Yang 2024: “community-dwelling older adults” → clarified as prefrail (n=100) + frail (n=100) groups; multicenter 13-community study; RCT is nested within larger observational design
  8. Yang 2024: “single-center” caveat → corrected to multicenter (13 communities, Xi’an, China)
  9. Buigues 2016: “community-dwelling” → nursing home residents (Valencia, Spain); 50 completers noted; exact p-values added (exhaustion p=0.002, grip p=0.04); overall frailty rate not significantly modified clarified
  10. Dosing section: “Buigues 2016: ~6 g/day; Yang 2024: dose not specified” → exact doses from PDFs added for all three trials
  11. All footnotes expanded with full author lists, full trial stats, NCT numbers, and PDF locations

Downstream pages to check (may cite PROMOTe or Yang 2024 claims):

[2026-05-07] verify | interventions/dietary/fmt

Pages verified: 1 (partial scope — Boehme 2021 and Chuang 2023 closed-access)

  • interventions/dietary/fmt.md — 6 factual corrections; AI-extraction banner removed; verified: true with scope

Sources read end-to-end (full PDF):

  • Smith 2017 (eLife, doi:10.7554/eLife.27014) — killifish heterochronic FMT; 26 pp.; locally available
  • Bárcena 2019 (Nat Med, doi:10.1038/s41591-019-0504-5) — progeroid mice FMT; 29 pp.; downloaded from OA repository
  • Chen 2020 (Aging, doi:10.18632/aging.102872) — centenarian→mouse FMT; 16 pp.; downloaded via PMC7138539
  • DeFilipp 2019 (NEJM, doi:10.1056/NEJMoa1910437) — ESBL case report; 5 pp.; downloaded

Sources unverifiable (closed-access):

  • Boehme 2021 (Nature Aging, doi:10.1038/s43587-021-00093-9) — not_oa; tagged no-fulltext-access in footnote and body
  • Chuang 2023 (Eur J Clin Microbiol Infect Dis, doi:10.1007/s10096-023-04644-3) — not_oa; tagged no-fulltext-access in footnote

Corrections made:

  1. Smith 2017: “near their median lifespan in germ-free conditions” → “near their median lifespan in captivity of 4–8 months” (fish are not germ-free; antibiotic depletion + recolonization protocol)
  2. Smith 2017: added specific lifespan-extension percentages to body (37% vs wt, 41% vs Omt, 21% vs Abx) with Logrank p-values
  3. Smith 2017: expanded footnote with control-group breakdown and quantitative outcomes
  4. Bárcena 2019: Lmna^G609G^ → Lmna^G609G/G609G^ (correct homozygous notation per paper)
  5. Bárcena 2019: added median lifespan numbers (160 vs 141 days, 13.5%, P=0.0029) and maximal survival data to body and footnote
  6. Chen 2020: “centenarian donors” (plural) → one 101-year-old donor + one 70-year-old elderly control (study had only 2 human donors total)
  7. Chen 2020: “young mice” → “11-month-old C57BL/6 mice” (recipients were middle-aged)
  8. Chen 2020: “n not extracted from abstract” → n=16 total (~8/group) per Methods
  9. Chen 2020: “no aged-mouse controls” → corrected (E group receiving elderly-donor FMT was the aged-control arm)
  10. Chen 2020: expanded SCFA genera (Roseburia, Faecalibacterium, Ruminococcus, Coprococcus) added
  11. Chen 2020: footnote updated with verified n, donor ages, and specific p-values ClinicalTrials.gov trial count (n=3, NCT05598112/NCT06649981/NCT06496412) confirmed via v2 API.

Downstream pages to review (main agent task):

[2026-05-07] verify | microbiome/bifidobacterium

Pages verified: 1 (partial scope — 3 sources unverifiable; see verified-scope)

  • microbiome/bifidobacterium.md — 7 factual corrections; AI-extraction banner removed; verified: true with scope

Sources read end-to-end (full PDF):

  • Claesson 2012 (Nature, doi:10.1038/nature11319) — n=178 elderly + 13 young controls; diet-microbiota correlation; PDF locally available
  • Mariat 2009 (BMC Microbiology, doi:10.1186/1471-2180-9-123) — n=62 (infants/adults/elderly); qPCR; PDF downloaded and read
  • Odamaki 2016 (BMC Microbiology, doi:10.1186/s12866-016-0708-5) — n=367; main article pp.1–6 read; PDF downloaded
  • Kim 2021 (J Gerontol A, doi:10.1093/gerona/glaa090) — n=63 randomized/53 analyzed; 12-wk RCT; PDF downloaded and read
  • Asaoka 2022 (J Alzheimers Dis, doi:10.3233/jad-220148) — n=130/115 analyzed; 24-wk RCT; PDF locally available; all 21 pages read
  • Falony 2006 (AEM, doi:10.1128/aem.01296-06) — in-vitro co-culture; PDF downloaded and read
  • Hopkins 2001 (Gut, doi:10.1136/gut.48.2.198) — n=10+7+5+4; polyphasic; PDF downloaded and read

Sources unverifiable:

  • Biagi 2016 (Curr Biol, doi:10.1016/j.cub.2016.04.016) — DOI lookup failed; no OA URL resolved; abstract only; tagged no-fulltext-access in body
  • Hopkins 2002 (J Med Microbiol, doi:10.1099/0022-1317-51-5-448) — not_oa; pre-existing no-fulltext-access maintained
  • Longhi 2024 (Beneficial Microbes, doi:10.1163/18762891-bja00013) — not_oa; pre-existing no-fulltext-access maintained

Corrections:

  1. Mariat 2009 cohort completely rewritten: n=91 “young adults, elderly, centenarians” → n=62 “infants n=21, adults n=21, elderly n=20; age 70–90 yr”; no centenarians in this study; F/B ratios (infants 0.4, adults 10.9, elderly 0.6) added; qPCR method specified; footnote n=91 → n=62
  2. Falony 2006 cross-feeding partners: fabricated species (Roseburia inulinivorans, Faecalibacterium prausnitzii, Eubacterium hallii) → actual paper species (Anaerostipes caccae DSM 14662 and Roseburia intestinalis DSM 14610); footnote updated to name correct co-culture partners
  3. Kim 2021 gut-bacteria outcome: fabricated “reduced fecal Enterobacteriaceae” → actual paper findings: reduced Eubacterium, Allisonella, Clostridiales, Prevotellaceae; analyzed n added (n=53, not 63); footnote updated
  4. Asaoka 2022 MCI diagnosis: “based on MMSE screening rather than clinical criteria” (backwards) → “DSM-5 clinical criteria plus MMSE 22–26 and CDR=0.5”; analyzed n=115 (not 130); VSRAD subgroup n=89 specified; VOI Z-score ≥1.0 cutoff added; footnote updated
  5. Hopkins 2001 description: “large-scale” removed; actual group sizes added (n=10 children, n=7 adults, n=5 elderly / n=4 analyzed, n=4 CDAD); inter-individual variability noted (2 of 4 elderly undetectable bifidobacteria)
  6. Biagi 2016 paragraph: verification-failure note added; “body Results section could not be verified” disclosed; no-fulltext-access tag added
  7. Footnote status updates: Mariat 2009 “PDF pending download” → “PDF locally verified”; Kim 2021 “PDF pending download” → “PDF locally verified”; Falony 2006 “PDF pending download” → “PDF locally verified”; Odamaki 2016 method and status updated

Downstream propagation needed (for main agent):

[2026-05-07] verify | model-organisms/loxodonta-africana

Pages verified: 1

  • model-organisms/loxodonta-africana.md — 6 corrections/enhancements; AI-extraction banner removed; verified: true with scope

Sources read end-to-end (full PDF):

  • Abegglen 2015 (JAMA) — doi:10.1001/jama.2015.13134
  • Sulak 2016 (eLife) — doi:10.7554/eLife.11994
  • Vazquez 2018 (Cell Reports) — doi:10.1016/j.celrep.2018.07.042
  • Nunney 2022 (Evol Appl) — doi:10.1111/eva.13383
  • Vollrath 2023 (TREE) — doi:10.1016/j.tree.2023.05.011

Corrections:

  1. Abegglen 2015 footnote — expanded: added in-vitro n’s (n=8 elephants, n=10 LFS, n=11 human controls); added 95% CIs for all apoptosis figures; clarified “early apoptosis (AV+PI−)”; added TP53 copy count claim attribution
  2. Apoptosis table — expanded: added 95% CIs for all cells; added “early apoptosis (AV+PI−)” column header; added method/n note below table
  3. Vazquez 2018 body text — expanded: added exact fold-induction values (8.18-fold DOX, 16.06-fold N3a); added Bak/Bax-dependent mechanism; added LIFR-independence finding (CHO cells); replaced general “induces apoptosis” with mechanistic specifics
  4. Vazquez 2018 footnote — rewritten: added quantitative fold-induction figures; added Bak/Bax-dependent + LIFR-independent detail; added timing of LIF6 refunctionalization (~59 mya)
  5. Nunney 2022 body text — expanded: added 157 aa ancestral truncation detail with pre-divergence timing; added dn/ds neutral-evolution finding; added duplication rate (~10^−6/generation); clarified Nunney’s concession that 1–2 copies may have been selectively favored; replaced vague “neutral drift” with full PAML/EMMC analysis context
  6. Nunney 2022 footnote — rewritten: added 157 aa ancestral truncation; added duplication rate; added nuance that 1–2 copies may have been favored
  7. Recency note — updated: removed schema-gap complaint (field now exists per R28); confirmed literature-checked-through: 2026-05-07

No corrections to: cancer mortality 4.81% (confirmed), TP53 count 20 (confirmed), Vollrath germline hypothesis (confirmed), Sulak retrogene expression/non-TF framing (confirmed)

literature-checked-through: 2026-05-07 (set in frontmatter)

Downstream pages that may cite corrected claims:

  • p53 — may cite Abegglen/Sulak apoptosis figures; check footnote CIs
  • genomic-instability — may cite elephant TP53 copy count; no numeric corrections needed (20 copies confirmed)
  • cancer — may cite 4.81% elephant cancer mortality; confirmed correct

[2026-05-07] verify | microbiome/gut-microbiome-aging-shifts

Pages verified: 1 (partial scope — Guan 2023 DOI corrected but full text not read; all other 6 sources read end-to-end)

  • microbiome/gut-microbiome-aging-shifts.md — 8 factual corrections; AI-extraction banner removed; verified: true with scope

Corrections:

  1. Guan 2023 DOI: 10.1039/d3mo00068a → 10.1039/d2mo00256f (fabricated DOI; correct DOI found via PubMed PMID:37204279)
  2. Bian 2017 n: “>1,000” → “1,095” (883 primary + 212 young soldier secondary cohort per Table 1)
  3. Bian 2017 age range: “ages 3–>100” → “ages 3–>94” (centenarian group is >94 per Table 1, not >100)
  4. Bian 2017 comparison range: “adults 30–100” → “adults 30–>94” (consistent with actual age groups)
  5. van Soest 2020 (NU-AGE) n: “>1,200” → “226” (this paper covers the Dutch NU-AGE cohort only, n=226 subjects with 452 fecal samples; the “>1,200” figure refers to the pan-European NU-AGE study, not this paper)
  6. Luan 2020 follow-up duration: “15-month” → “20-month follow-up; fecal samples collected over 15 months” (Results section states survey lasted 20 months; abstract states samples collected for 15 months — these are different quantities; wiki had confused sample-collection window with total follow-up)
  7. Wilmanski 2021 uniqueness metric definition added: “minimum Bray-Curtis dissimilarity of each individual from their nearest neighbor” (not previously specified); survival analysis correctly scoped to MrOS longitudinal cohort only (N=706 community-dwelling; 85+ subgroup N=256); Arivale and AGP are cross-sectional only
  8. Wu 2019 n: “cross-sectional” (no n) → “65 recruited; 59 qualified stool samples; centenarians n=21, elderly n=25, young n=19” (per Results and Table 1)

Additional details added per verified source:

  • Claesson 2012: community n=83/day hospital n=20/rehab n=15/long-stay n=60 breakdown added; Procrustes Monte Carlo P<0.0001 for diet-microbiota co-separation added; phylum-direction clarified (long-stay higher Bacteroidetes, community higher Firmicutes)
  • Luan 2020: all 10 differentially changing species named in footnote; ANOSIM R=0.10, P=0.02 added
  • Wu 2019: Eubacterium rectale co-depletion added; enriched species (Methanobrevibacter smithii, Bifidobacterium adolescentis) added; functional pathway detail added
  • van Soest 2020: study design clarified (RCT but dietary intervention did not affect microbiota; analyzed cross-sectionally); Eubacterium rectale added to correlated taxa

Schema gap noted by seeder (for round-close decision, not fixed here):

  • type: process lacks literature-checked-through field; this page covers fast-moving literature. Consider adding literature-checked-through: to the type:process schema in CLAUDE.md (same pattern as type:compound/intervention).

ROADMAP note (for round-close, not fixed here):

  • Seeder used [x] rather than deleting the ROADMAP line per project convention; flag for round-close cleanup.

Downstream propagation needed (for main agent):

  • dysbiosis hallmark page — if it cites any of the corrected claims (NU-AGE n, Bian age range, Luan follow-up), update
  • akkermansia-muciniphila — if it cites Luan 2020 for the pre-death decline claim, verify the 8-species list is accurate
  • Any page citing guan-2023-butyrate-myoblast or the old DOI 10.1039/d3mo00068a — update DOI

[2026-05-07] verify | pathways/lps-tlr4-nfkb

Pages verified: 1 (partial scope — see verified-scope in frontmatter)

  • pathways/lps-tlr4-nfkb.md — 6 factual corrections; banner removed; verified: true with scope

Corrections:

  1. “2–5-fold above young adult levels in older adults [^kim2016]” → removed fabricated human fold-change; Kim 2016 is a mouse study (C57BL/6J) with no stated fold-change number; claim rewritten to reference elevated LPS in aged mice with gap tag.
  2. Firmicutes claim reversed: wiki said “Firmicutes reduction” attributed to Kim 2016, but Kim 2016 shows Firmicutes and Actinobacteria INCREASED in aged mice, Bacteroidetes DECREASED — Firmicutes/Bacteroidetes ratio elevated. Section rewritten to reflect actual paper findings with species-specificity caveat.
  3. Bailey 2019 body claim “elevated baseline TNFα and IL-6 / reduced NF-κB phosphorylation kinetics” → corrected: Bailey 2019 measures TNFα only (not IL-6), uses functional TNFα output as proxy (not direct NF-κB phosphorylation measurement). Precise statistics added: 2.5%/yr baseline increase (p=0.02), 1.9%/yr ΔTNFα decrease post-LPS (p=0.05).
  4. LBP/sCD14 paragraph citation [^bailey2019] removed as fabricated attribution: Bailey 2019 contains no LBP or sCD14 data. Tagged unsourced with candidate replacement source (Stehle 2012, cited in Kim 2016).
  5. TAK-242 table entry incorrectly attributed the ACCESS trial (n=1,961) to TAK-242 → corrected: ACCESS trial is eritoran’s Phase 3 trial. TAK-242 had a separate Phase 3 (LOTUS trial, Shieh 2011, n=274). Table note updated.
  6. Matsunaga 2010 footnote: year noted as 2011 print / 2010 epub; actual paper contribution clarified as adaptor-interference mechanism (TIRAP/TRAM disruption) — Cys747 identification was Takashima 2009, which Matsunaga 2010 confirmed.

Sources read end-to-end from PDF:

  • Kim 2016 (BMC Microbiology, doi:10.1186/s12866-016-0625-7): full 9-page PDF; n=8/group confirmed; Firmicutes increase finding confirmed; p16 and NF-κB Western blot results confirmed
  • Opal 2013 (JAMA, doi:10.1001/jama.2013.2194): full PDF; ACCESS trial: eritoran n=1304, placebo n=657, MITT n=1961; 28-day mortality 28.1% vs 26.9%, p=0.59 — all confirmed
  • Bailey 2019 (Aging Clin Exp Res, PMC6504629): full text via PMC efetch; TNFα-only outcome; no LBP/sCD14 measurements; quantitative per-year changes confirmed

Sources unverifiable (closed-access, abstract only):

  • Matsunaga 2010 (Mol Pharmacol) — not_oa; abstract via PubMed (PMID 20881006) confirms adaptor-interference mechanism; Cys747 attribution to Takashima 2009 confirmed
  • Li 2005 (J Mol Med) — not_oa; review; claims unchecked
  • Nagpal 2016 (Beneficial Microbes) — not_oa; review; claims unchecked
  • Yao 2023 (Inflammopharmacology) — not_oa; TAK-242 aged-rodent neuroinflammation claim unchecked

Downstream propagation needed (for main agent):

  • Any entity page citing lps-tlr4-nfkb for the “2–5-fold LPS elevation” human claim — remove or re-source
  • Any page citing lps-tlr4-nfkb for the Firmicutes aging direction — add model-dependence caveat

[2026-05-07] verify | processes/gut-barrier

Pages verified: 1 (partial scope — see verified-scope in frontmatter)

  • processes/gut-barrier.md — 10 corrections/additions; AI-extraction banner removed; verified: true with scope

Corrections:

  1. Thevaranjan 2017 mouse age: “24 mo” → “18–22 mo” (fabricated by seeder)
  2. Thevaranjan 2017 group sizes: “n ~10–20/group” → “n=4–12 per experiment” (multiple assays with different n)
  3. Thevaranjan 2017 bacterial product: “systemic LPS” → MDP (muramyl dipeptide) is the measured bacterial product; LPS was ex vivo stimulant only
  4. Thevaranjan 2017 cytokines: TNF-α only → TNF + IL-6 both reported
  5. Thevaranjan 2017 anti-TNF: misleading inference removed; paper states anti-TNF had no effect on permeability
  6. Thevaranjan 2017 2018 correction noted in footnote
  7. Depommier 2019: precise ± values added; plasma LPS secondary endpoint confirmed reduced; L:M ratio not measured (corrects “barrier-specific endpoints” claim)
  8. Wu 2026: DOI confirmed (10.1111/acel.70533); author list corrected; PMCID added; FMT-to-GF framing flagged unsourced (not in abstract)
  9. Butyrate mechanism: Peng 2009 added for AMPK-mediated TJ assembly; HDAC/H3 claim remains unsourced
  10. Shimizu 2022: absolute HD5 means added; Pearson method noted; DOSANCO cohort identified

Unverifiable: Wu 2026 full text (not in archive; PMC full text not read — FMT design needs confirmation)

Downstream propagation needed (for main agent):

  • Any entity page citing thevaranjan2017 with “24 mo” or “LPS translocation” framing
  • akkermansia-muciniphila compound/microbe page if it cites Depommier 2019

[2026-05-07] verify | pathways/fgf-signaling

Pages verified: 1 (partial scope — see verified-scope in frontmatter)

  • pathways/fgf-signaling.md — 5 factual corrections; banner removed; verified-scope documents Wedemeyer 2025 + Kuro-o 1997 + Yamada 2017 closed-access/unresolved constraints

Sources read (PDF end-to-end):

  • Zhang 2012 (eLife, doi:10.7554/elife.00065): lifespan figure corrected from “~36–40%” to sex-stratified (~30% males / ~39% females, combined median 28.1→38.2 mo); n counts added (Tg=77, WT=67); mechanism clarified (JAK2/STAT5/IGF-1; AMP kinase + mTOR + sirtuin markers NOT elevated); HR=0.22 [0.15–0.34], p=2.7×10⁻¹² added; strain confirmed C57Bl/6J
  • Chang 2024 (EMBO Journal, PMC10897314): ISR mechanism corrected to include PKR as eIF2α kinase upstream of ATF4; cellular source confirmed BAT-specific; UCP1-independence confirmed
  • Youm 2025 (Nature Aging, PMC12003152): cellular sources corrected — both TEC-derived and adipocyte-derived FGF21 are independently protective; hepatocyte-derived endocrine FGF21 (5×) does NOT affect thymus; βKlotho ablation in Foxn1⁺ TECs confirmed; n=6–13/group added

Sources closed-access/unresolvable:

  • Kuro-o 1997 (Nature) — not_oa; 61-day lifespan figure unverified; noted in footnote
  • Wedemeyer 2025 (Nature Aging) — not_oa; body softened to title-only claims; tagged no-fulltext-access
  • Yamada 2017 (Bone) — PMC listed but download URL resolution failed; review paper; unverified

Druggability spot-check: Open Targets 2026-05-07 — FGF21 + FGFR1 both tier 1 (max); aging-context tier 2 confirmed correct and documented inline.

Downstream pages to update (for main agent):

  • fgf21 — if seeded: needs sex-stratified lifespan figures and PKR-upstream ISR mechanism
  • klotho — if seeded: Kuro-o 1997 citation should note no-fulltext-access

[2026-05-07] verify | allele-selective-oligonucleotides

Pages verified: 1

  • interventions/pharmacological/allele-selective-oligonucleotides.md — 5 corrections applied; AI-extraction banner removed; verified: true

Corrections:

  1. Nusinersen ENDEAR n: “121 (73+37)” → “122 enrolled; final analysis n=110 (73+37)” per CT.gov NCT02193074 + NEJM abstract (73+37=110, not 121)
  2. Inotersen NEURO-TTR footnote: “n=173 (112 inotersen, 61 placebo)” → “173 randomized; 172 received at least one dose (112 inotersen, 60 placebo)” per NEJM abstract (PMID 29972757) + CT.gov NCT01737398
  3. Inclisiran ORION-9 n: “490 (242+248)” → “482 randomized 1:1” per NEJM abstract (PMID 32197277) + CT.gov NCT03397121
  4. Bragg 2026 DOI resolved: PMID 41849580 → doi:10.1126/scitranslmed.adv0702 (Sci Transl Med 2026 Mar 18;18(841):eadv0702); full author list confirmed; no-fulltext-access removed
  5. Maeda 2026 DOI resolved: PMID 41552385 → doi:10.1016/j.omtn.2025.102802 (Mol Ther Nucleic Acids 2026 Mar 12;37(1):102802); PMCID PMC12804371 added; “serinol nucleic acid” corrected to “acyclic serinol nucleic acid (SNA)”; disease models (SBMA, SCA3) added to body text

Confirmed correct (no change):

  • Tofersen VALOR: ALSFRS-R difference +1.2 points, P=0.97, primary endpoint NOT met — confirmed against NEJM abstract (PMID 36129998)
  • Patisiran APOLLO: mNIS+7 −34.0 pts, P<0.001, n=225 (148+77) — confirmed
  • Vutrisiran HELIOS-A: P=3.54×10⁻¹², n=164 (122+42 + 77 external placebo) — confirmed
  • Givosiran ENVISION: 74% attack rate reduction, P<0.001, n=94 (48+46) — confirmed
  • Lumasiran ILLUMINATE-A: −53.5 pp, P<0.001, n=39 (26+13) — confirmed
  • clinical-trials-active: null — ClinicalTrials.gov API reconfirmed (0 results for allele-selective ASO/siRNA, GNAQ siRNA, DNMT3A ASO)

Sources: all 8 pivotal-trial PDFs pending/not_oa; endpoint claims verified via PubMed efetch abstracts only; 2026 papers Bragg + Maeda verified via PubMed efetch abstracts

Supersession candidates (R25): none — no 2024–2026 Phase 3 RCT or meta-analysis on allele-selective oligonucleotide aging applications; literature-checked-through: 2026-05-07 confirmed Downstream pages to update: none (no other wiki pages cite this intervention page yet per seeder log)

[2026-05-07] verify | model-organisms/balaena-mysticetus

Pages verified: 1

  • model-organisms/balaena-mysticetus.md — corrections: 4 footnote errors, 1 new gap tag, banner removed; verified-scope documents George 1999 closed-access constraint

Sources read (PDF):

  • Keane 2015 (Cell Reports, PMC4536333): full 12-page PDF read; CDKN2 duplication confirmed ABSENT; PCNA duplication, ERCC1/ERCC3 positive selection, FOXO3 positive selection all confirmed; volume corrected 9→10; author count corrected 29→30
  • Firsanov 2025 (Nature, PMC12711569): full 10-page PDF read (pages 1-10); CIRBP mechanism confirmed; “rather than relying on additional tumour suppressor genes” exact quote confirmed in abstract; bowhead requires fewer oncogenic hits confirmed; NHEJ + HR enhancement confirmed; Drosophila lifespan extension confirmed
  • Swovick 2020 (MCP): article number corrected 100004→100041; year corrected 2020→2021 (published 2021)
  • Seim 2014 (Aging): issue/pages corrected 6(12):1080-1092→6(10):879-899

Sources closed-access (unverifiable from PDF):

  • George 1999 (Can. J. Zool.) — not_oa; 211-year estimate corroborated by Keane 2015 PDF and AnAge ID 02275 (confirmed online)

Recency check (R25/R28): PubMed search for bowhead whale longevity/DNA repair 2025-2026 found 2 papers: PMID 41162698 (Firsanov 2025 — already the primary source on page) and PMID 41960904 (Paniagua 2026 — confirmed commentary only, no primary findings). No supersession detected.

Downstream pages to update (for main agent):

  • None identified — this is a model-organism page with no other wiki pages currently citing it by wikilink (no downstream propagation needed for footnote corrections)

[2026-05-07] ingest | allele-selective-oligonucleotides

  • added: interventions/pharmacological/allele-selective-oligonucleotides.md
  • added class to: frameworks/intervention-classes.md (new class allele-selective-degradation / RNA-targeted-knockdown with sub-values rnase-h-recruitment, rna-interference, splice-modulation)
  • entity type: type: intervention | mode: pharmacological
  • sections: class definition (ASO vs siRNA vs splice-modulation), clinical landscape (9 FDA-approved drugs with verified pivotal trial outcomes), allele-selectivity technical primer, aging-relevance (GNAQ Q209L cherry angioma, CHIP, splice dysregulation), delivery bottlenecks, safety profile, limitations/gaps
  • FDA-approved drugs documented: patisiran, vutrisiran, givosiran, lumasiran, inclisiran (siRNA); inotersen, nusinersen, tofersen, eteplirsen/casimersen/viltolarsen/golodirsen (ASOs); mipomersen (withdrawn, historical)
  • key distinction established: current approved drugs are NOT allele-selective (they silence wild-type gene transcripts in inherited disease); allele-selective (SNV-discriminating) programs are research-stage only (HD/polyglutamine programs; no aging-specific programs)
  • pivotal trial DOIs verified via PubMed efetch + DOI lookup (all at 100th citation percentile in archive): 10.1056/NEJMoa1716153 (patisiran), 10.1056/NEJMoa1716793 (inotersen), 10.1056/NEJMoa1702752 (nusinersen), 10.1056/NEJMoa2204705 (tofersen), 10.1056/NEJMoa1913147 (givosiran), 10.1056/NEJMoa2021712 (lumasiran), 10.1056/NEJMoa1913805 (inclisiran), 10.1080/13506129.2022.2091985 (vutrisiran)
  • tofersen nuance: VALOR trial primary endpoint NOT met (ALSFRS-R difference P=0.97); FDA accelerated approval based on CSF SOD1 + NfL biomarker evidence — documented accurately (not framed as trial success)
  • recency search (R25): searched 2021-2026 for allele-selective oligonucleotide + aging/somatic mutation (3 hits), GNAQ Q209L siRNA (0 hits), DNMT3A allele-specific oligonucleotide CHIP (0 hits), FDA oligonucleotide approvals 2024 (6 hits — two narrative reviews, one bioRxiv preprint on SOD1 siRNA); two 2026 papers on HD/polyglutamine allele-selective programs integrated (PMID 41849580, 41552385); no programs superseding the training-era landscape identified
  • ClinicalTrials.gov search for allele-selective ASO trials: 0 active trials returned (verified via API)
  • R16 compliance: hallmarks link to [[genomic-instability]] (page exists); translation-gap: phase-3-rct-needed (aging indication); next-experiment: populated
  • implicit stubs created: none (all cross-linked pages already exist)
  • gaps surfaced: no allele-selective oligonucleotide program for any aging somatic mutation in clinical development; HSC delivery for CHIP application is the dominant translational bottleneck; no topical/intralesional siRNA program for cherry angioma identified

[2026-05-07] ingest | balaena-mysticetus

  • added: model-organisms/balaena-mysticetus.md
  • entity: Balaena mysticetus (bowhead whale), type: model-organism
  • primary sources cited (DOIs verified via PubMed efetch + Crossref + DOI lookup; PDFs pending):
    • 10.1139/z99-015 (George 1999, Can. J. Zool. — aspartic acid racemization; max 211 yr; confirmed via Crossref; not_oa)
    • 10.1016/j.celrep.2014.12.008 (Keane 2015, Cell Reports — bowhead genome; PCNA duplication, ERCC1/ERCC3 positive selection, FOXO3 positive selection; confirmed PMC4536333; gold OA)
    • 10.1038/s41586-025-09694-5 (Firsanov 2025, Nature — CIRBP enhanced DNA repair; confirmed PMID 41162698, PMC12711569; green OA)
    • 10.18632/aging.100699 (Seim 2014, Aging — bowhead transcriptome; confirmed PMID 25411232; green OA)
    • 10.1074/mcp.RA120.002301 (Swovick 2020, Mol Cell Proteomics — protein turnover rates; confirmed PMID 33639418)
    • 10.1016/j.cell.2023.05.002 (Tyshkovskiy 2023, Cell — 41-species longevity signatures; confirmed PMID 37269831; green OA)
    • 10.7554/eLife.51089 (Kowalczyk 2020, eLife — pan-mammalian constraint analysis; confirmed PMID 32043462; gold OA)
  • CRITICAL SEEDER CORRECTION: seeder brief claimed “CDKN2 family duplication” as primary cancer-resistance mechanism. PubMed search for “bowhead whale CDKN2 duplication” = 0 results. Keane 2015 genome paper does NOT report CDKN2/INK4 family duplication. Firsanov 2025 explicitly states bowhead does NOT rely on tumor suppressor gene expansion. CDKN2 claim tagged unsourced and flagged in body. Correction prominently documented in auto-extraction banner and body.
  • recency search (model-organism; non-mandatory R25 but performed given comparative-biology importance): PubMed “bowhead whale longevity genome” 2020-2026 — 7 hits triaged; Firsanov 2025 (Nature) is the key superseding mechanistic paper — integrated as primary source. No meta-analyses or RCTs (not applicable for model-organism page type).
  • implicit stubs confirmed already resolved: loxodonta-africana (seeded same day), heterocephalus-glaber (exists), homo-sapiens (exists), mus-musculus (exists)
  • implicit stubs still needed: dna-damage-response (DDR pathway page), cancer (phenotype page)
  • schema note: type: model-organism does not currently include literature-checked-through: field (escalated in loxodonta-africana log; same gap applies here)
  • AnAge ID 02275 confirmed via AnAge query for Balaena mysticetus

[2026-05-07] ingest | loxodonta-africana

  • added: model-organisms/loxodonta-africana.md
  • entity: Loxodonta africana (African bush elephant), type: model-organism
  • primary sources cited (DOIs verified via PubMed efetch + Crossref; PDFs pending in archive):
    • 10.1001/jama.2015.13134 (Abegglen 2015, JAMA — cancer mortality + apoptosis hypersensitivity data)
    • 10.7554/eLife.11994 (Sulak 2016, eLife — 20 TP53 copies; retrogene expression)
    • 10.1016/j.celrep.2018.07.042 (Vazquez 2018, Cell Reports — LIF6 zombie gene, mitochondrial apoptosis)
    • 10.1016/j.tree.2011.01.002 (Caulin & Maley 2011, TREE — Peto’s paradox framework; not_oa)
    • 10.1111/eva.13383 (Nunney 2022, Evol Appl — mechanistic challenge: most retrogenes truncated, no positive selection)
    • 10.1016/j.tree.2023.05.011 (Vollrath 2023, TREE — germline protection alternative hypothesis)
    • 10.1093/molbev/msac149 (Padariya 2022, Mol Biol Evol — MDM2-escape isoforms; in-silico)
    • 10.1530/REP-25-0254 (Moore 2025, Reproduction — reproduction vs cancer function commentary)
  • recency search: 7 post-2021 papers triaged; Nunney 2022 and Vollrath 2023 directly challenge training-era TP53 causal story — framed explicitly in body
  • gaps surfaced: TP53 retrogene functional count contested; LIF6 mechanism in-vivo unverified; telomere and senescence biology unstudied; genome-to-human similarity not confirmed; schema gap: type: model-organism lacks literature-checked-through: field
  • implicit stubs created: balaena-mysticetus, intervention-by-target-immunogenicity, cancer, cherry-angioma

[2026-05-07] ingest | intervention-by-target-immunogenicity

  • added: frameworks/intervention-by-target-immunogenicity.md
  • source: promoted from “Decision framework: target immunogenicity dictates intervention choice” section on hallmarks/disabled-adaptive-immunity.md
  • rationale: framework logic applies wiki-wide (not only to immune aging); standalone MOC enables cross-hallmark navigation by immunogenicity axis
  • sections: opening principle, four-tier table, Tier 2 depth (CHIP), Tier 1 depth (uPAR/GPNMB), comparative biology (NMR/elephant/bowhead), limitations, research priority implications, cross-references
  • implicit stubs created: [[model-organisms/loxodonta-africana]], [[model-organisms/balaena-mysticetus]], [[interventions/pharmacological/allele-selective-oligonucleotides]]
  • recency check (light, framework page): PubMed 2023-2026 searches for “CAR-T senolytic aging” (28 hits), “neoantigen immunogenicity aging” (81 hits), “immunogenicity senescent cells” (9 hits) — no superseding meta-analysis or framework-contradicting RCT identified; framework consistent with active literature direction
  • no schema violations; type: framework does not carry verified discipline per CLAUDE.md
  • gaps surfaced: elephant + bowhead whale model-organism pages missing; allele-selective-oligonucleotides intervention page missing

[2026-05-07] verify | disabled-adaptive-immunity

Pages verified: 1

  • hallmarks/disabled-adaptive-immunity.md — corrections: 7 (see below)

Sources checked:

  • doi:10.1111/acel.13028 (Fahy 2019, Aging Cell) — primary PDF downloaded and verified
  • doi:10.1038/s41586-020-2403-9 (Amor 2020, Nature) — primary PDF downloaded and verified
  • doi:10.1016/S2666-7568(21)00062-3 (Mannick 2021, Lancet Healthy Longevity) — primary PDF downloaded and verified
  • doi:10.1126/scitranslmed.3009892 (Mannick 2014, Sci Transl Med) — not_oa; verified via PubMed efetch abstract PMID 25540326
  • doi:10.4049/jimmunol.175.5.2982 (Heng 2005, J Immunol) — DOI confirmed via Crossref; PMID 16116185; OA pending
  • doi:10.1242/dev.103614 (Bredenkamp 2014, Development) — DOI corrected from dev.115899; confirmed PMID 24715454 via PubMed
  • doi:10.1038/s41418-026-01724-7 (Genah 2026, Cell Death Differ) — real paper confirmed PMID 41888548 via PubMed; mechanism verified vs abstract
  • doi:10.1080/21645515.2025.2479229 (Zhang 2025 GPNMB nanovaccine) — DOI resolved from PMID 40088037
  • doi:10.1016/j.biopha.2024.117433 (Cao 2024 GPNMB DC vaccine) — DOI resolved from PMID 39260327
  • doi:10.1038/s41590-017-0006-x (Nikolich-Žugich 2018) — not_oa; review citation only
  • doi:10.1038/s41577-019-0180-1 (Goronzy 2019) — download failed; review citation only
  • doi:10.1093/gerona/glu057 (Franceschi 2014) — not_oa; review citation only
  • doi:10.1126/science.adf8531 (Dai 2024 ZEB2/ABC) — pending download; qualitative-only claims; consistent with PubMed record
  • doi:10.1172/JCI146071 (Sato 2022 CD153/TLT) — pending download; qualitative-only claims; consistent with PubMed record

Corrections made:

  1. Fahy 2019 enrollment: “enrolled 9 healthy males” → “enrolled 10 healthy males (9 completers; 1 dropout at ~1 month)” — matches paper methods section
  2. Fahy 2019 thymic MRI response: “regeneration in all 9 subjects” → “in 7 of 9 completing subjects (2 non-responders had abnormally low baseline TFFF)” — matches Figure 3 and paper text
  3. Fahy 2019 rate framing: added baseline −1.6 yr/yr (months 0–9) and the −2.5-yr change vs no-treatment framing alongside the −1.5-yr vs baseline figure; GrimAge 2-year decrease figure added — all confirmed from Table 1 and abstract
  4. Amor 2020 lifespan claim: removed “Extended physiological lifespan when administered to old wild-type mice” (not in main paper text; only lung adenocarcinoma survival and liver fibrosis reversal are in main text) — replaced with note and needs-replication tag
  5. Bredenkamp 2014 DOI: corrected dev.115899 → dev.103614 (confirmed PMID 24715454 via PubMed efetch)
  6. GPNMB zhang2025 DOI: “doi:not-confirmed-verify” → confirmed doi:10.1080/21645515.2025.2479229 via PMID 40088037
  7. GPNMB cao2024 DOI: “doi:not-confirmed-verify” → confirmed doi:10.1016/j.biopha.2024.117433 via PMID 39260327

Additional updates (non-corrections):

  • Genah 2026 footnote expanded with full author list, confirmed PMID, institution, mechanism chain (UPR detail added), and paper scope
  • All footnotes updated with PMID/PMCID fields and verification method
  • Mannick 2021 footnote enriched with exact Phase 2b and Phase 3 n/outcome numbers from PDF
  • Heng 2005 footnote clarified: ETP number (not reconstitution potential) was restored; chemo-model results added
  • AI-extraction banner removed

Unverifiable claims:

  • Nikolich-Žugich 2018 CMV memory inflation “10–50% of total CD8+ compartment” — not_oa; no-fulltext-access preserved on that claim
  • Mannick 2014 exact n and age range (“~70 years old”) — abstract confirms elderly volunteers but does not specify n; closed-access; wiki text framing is consistent with abstract
  • Dai 2024, Sato 2022, Zhang 2023 (Blood Adv) — qualitative mechanism claims only; no primary quantitative claims extracted to this hallmark page

Recency check (R25):

  • PubMed 2024–2026 search for “TRIIM thymic regeneration”: no superseding studies found
  • PubMed 2024–2026 search for “mTOR inhibitor immune aging elderly randomized”: no new Phase 3 results found
  • literature-checked-through: 2026-05-07 (confirmed, not updated)

Schema-gap escalations (re-stated for main agent):

  1. introduced: proposed is not in the CLAUDE.md type: hallmark enum (currently 2013 | 2023). Recommend adding proposed as a valid value to capture candidate-hallmark status.
  2. literature-checked-through: field is formally defined only on type: compound, type: intervention, type: biomarker, type: hypothesis. This hallmark page uses it because of high-velocity biotech intervention content. Recommend adding to type: hallmark schema.

Final state: verified: true

[2026-05-07] ingest | disabled-adaptive-immunity (ad-hoc seed, user-requested)

  • added: hallmarks/disabled-adaptive-immunity.md
  • entity type: proposed candidate hallmark (type: hallmark, introduced: proposed — schema extension)
  • key sections: why-a-hallmark-candidate (case for and against), mechanistic-core (thymic-involution, T-cell repertoire contraction, B-cell remodeling, NK decline, HSC myeloid bias, macrophage dysfunction), functional consequences, biotech intervention landscape (5 mechanism classes: thymic regeneration, CAR-T senolytics, NK/IL-15 axis, mTOR inhibition, HSC rejuvenation), causal-graph integration, knowledge gaps
  • canonical IDs pulled: none (hallmark page; no UniProt/PubChem applicable)
  • primary-source DOIs cited (12 footnotes): 10.1038/s41590-017-0006-x, 10.1038/s41577-019-0180-1, 10.1093/gerona/glu057, 10.1111/acel.13028, 10.1242/dev.115899, 10.4049/jimmunol.175.5.2982, 10.1038/s41586-020-2403-9, 10.1038/s41418-026-01724-7, 10.1126/scitranslmed.3009892, 10.1016/S2666-7568(21)00062-3, 10.1126/science.adf8531, 10.1172/JCI146071
  • recency search: PubMed 2023–2026 searched for adaptive immunity + aging hallmark proposals (2 hits, neither specifically proposing hallmark); thymic regeneration clinical trials (TRIIM-X NCT04375657 confirmed recruiting May 2025); CAR-T senolytics (Amor 2020 Nature, 10 recent PMIDs retrieved); GPNMB vaccine papers (2024–2025, PMIDs 40088037 + 39260327); ZEB2/ABC paper (2024 Science, PMID 38271512); Mannick RTB101 Phase 3 failure (PMID 33977284). literature-checked-through: 2026-05-07
  • schema gaps surfaced: (1) introduced: proposed not in CLAUDE.md hallmark-type enum; (2) literature-checked-through: field used on type:hallmark but not defined in schema — both flagged for CLAUDE.md update
  • implicit stubs created: il-7-signaling, clonal-hematopoiesis (may exist — check), cancer (exists), apoptosenes (check)
  • programs verified as real: TRIIM-X NCT04375657 (recruiting); uPAR CAR-T (Amor 2020 Nature — confirmed); N-803/Anktiva FDA-approved 2024 (confirmed); RTB101 Phase 3 failure (Mannick 2021 Lancet Healthy Longevity — confirmed); ZEB2/ABC (2024 Science — confirmed); nitazoxanide/FOXN1 (Genah 2026 Cell Death Diff — confirmed)
  • programs that did NOT check out: “BPX-01” as mentioned in seeder brief — not found in any search; dropped
  • DOIs for zhang2025 and cao2024 (GPNMB vaccine papers) confirmed via PMID but DOIs not confirmed via archive; flagged as doi:not-confirmed-verify — verifier must resolve

[2026-05-07] verify | cherry-angioma

  • page verified: phenotypes/cherry-angioma.md
  • sources checked (8 citations):
    • Liau 2019 (PMID 31189994, closed-access): verified via PubMed + Europe PMC abstract; 82% mutation rate confirmed; GNA14 41% all cases / 47% classic cases (corrected from “~35–40%”)
    • Klebanov 2019 (PMC6440195, green OA): verified via PMC full text; n=10 confirmed, 5 specific mutations confirmed; fabricated functional-experiment claim removed (“GNAQ intrabody LoF suppresses proliferation in cell culture” — paper contains no cell culture experiments)
    • Nakashima 2010 (primary PDF): n corrected 8→7 per group (7 senile hemangioma + 7 controls)
    • Kumar 2021 (PMID 33911297): verified via PubMed + Europe PMC abstract; n=250, 33.2%, mean age 67.87±7.29 confirmed (minor rounding in body text preserved)
    • Pastor-Tomás 2023 (primary PDF): “most common vascular tumors” confirmed; associations table verified; no numeric corrections needed
    • Bujoreanu 2025 (primary PDF): n added (60); biologic-class direction corrected — wiki said “TNF-α inhibitors associated with more angioma formation than IL-23 inhibitors” but paper shows the opposite: IL-23 inhibitors had higher counts (6.94 vs 5.06); between-class difference p=0.798 (ns) added
    • Askari 2013 (PMID 23370299, closed-access): verified via PubMed abstract; n added (72+268); prolactin direction corrected — wiki said “elevated” but paper shows prolactin was lower in angioma cases (10.13 vs 13.13, p=0.0096); CCL2 elevation (203.5 vs 187.1 pg/ml, p=0.035) confirmed
    • Zhao 2023 (PMID 37701250, closed-access): verified via PubMed abstract; case confirmed as 68-yr-old male with recurrent strokes; outcome details added; IVLBCL literature count updated (≥21 cases by 2024)
  • StatPearls [^qadeer-statpearls]: unverifiable (HTTP 403); specific “75% by age 70” figure remains unsourced
  • corrections: 6 substantive changes (Klebanov fabricated claim, Nakashima n, Bujoreanu direction+n, Askari n+prolactin direction, GNA14 percentage precision, Kumar age precision)
  • recency check: no superseding GNAQ/GNA11/GNA14 mutation study 2023–2026; BMC Geriatrics 2025 (PMID 40188076, n=552) flagged as new epidemiology (not superseding)
  • literature-checked-through: 2026-05-07

[2026-05-07] ingest | cherry-angioma (ad-hoc seed, user-requested)

  • added: phenotypes/cherry-angioma.md
  • gaps surfaced: no-fulltext-access (all 8 cited papers pending download or closed-access); unsourced (age-stratified prevalence figures are textbook-derived, no indexed primary epidemiology study found); no-mechanism (trunk distribution, mutation timing vs clonal expansion, SASP/senescence link, miR-424 + GNA14 interaction); needs-replication (bFGF/FGFR1 axis in cherry angioma; MEK inhibitor therapeutic angle); needs-canonical-id (ICD-10-CM D18.01 is best available code — verify cherry-angioma–specific billing code)
  • implicit stubs created: none new (skin-aging page exists; hallmark pages exist)
  • recency search: PubMed 2021–2026 sweep performed; key recent papers integrated (Bujoreanu 2025, Liau 2019 Modern Pathology, Klebanov 2019 JAMA Derm, Pastor-Tomás 2023 Actas Dermo); no RCTs or meta-analyses exist in this space

[2026-05-07] R32 — DNA-repair pathway completion (DONE; 15/15 pages seeded + verified)

Round 32 status: COMPLETE. Targets the genomic-instability intervention asymmetry flagged in the 2026-05-07 lint pass: 23 mech proteins / only 7 interventions despite mechanism-deep pathway coverage.

Pages seeded + verified across three sub-batches (15):

R32a — BER cluster (5):

  • molecules/proteins/pcna.md — replication clamp + multi-pathway hub (BER/NER/MMR/TLS); homotrimeric ring; K164 modifications (mUb/polyUb/SUMO2)
  • molecules/proteins/xrcc1.md — BER scaffold; SCAR26 disease (Hoch 2017); BRCT1 makes direct protein-protein contact with auto-PARylated PARP1 (NOT generic PAR-chain reader)
  • molecules/proteins/ape1.md — AP endonuclease + REF-1 redox factor; conditional-KO mouse (Li & Wilson 2018) shows premature aging
  • molecules/proteins/mutyh.md — DNA glycosylase for A:8-oxoG mispairs; MUTYH-associated polyposis (Al-Tassan 2002)
  • molecules/proteins/polb.md — short-patch BER polymerase + dRP lyase; mtBER role newly established (Baptiste 2021 contradicts prior POLγ-exclusive model)

R32b — HR/NER + interventions (5):

  • molecules/proteins/lig1.md — replicative + long-patch BER ligase (NOT short-patch — Levin 2000 / Pascucci 1999); Bentley 1996 LIG1-/- is perinatal lethal hematopoietic-specific (NOT embryonic)
  • molecules/proteins/rpa.md — heterotrimer (RPA70/32/14) ssDNA coater; Toledo 2013 RPA exhaustion model; trimerization via parallel α-helices (NOT OB-fold interface)
  • molecules/proteins/xpf.md — NER 5’ incision endonuclease (with ERCC1); XFE progeroid syndrome (Niedernhofer 2006); DR rescue (Vermeij 2016)
  • molecules/proteins/brca2.md — RAD51 loader for HR; FANCD1; BRCA2 directs RAD51 to ssDNA but does NOT directly displace RPA (Jensen 2010 Fig. 1d pull-down negative)
  • interventions/pharmacological/parp-inhibitors.md — class page; oncology-established + aging-speculative; CH-induction safety signal documented (Marshall 2024 / Nuttall Musson 2024)

R32c — MMR/DDR + DNA-PKcs (5):

  • molecules/proteins/msh2.md — MutSα + MutSβ partner; ~40% Lynch syndrome cases; Lamers 2000 MutS structure has 6 regions (not 5)
  • molecules/proteins/mlh1.md — MutLα/β/γ partner; ~50% Lynch syndrome; Herman 1998 (NOT Kane 1997 — Kane is 10th author) for sporadic MSI-H methylation; MutLγ “backup MMR” claim was inverted (Lipkin 2002 actually says NOT required for somatic MMR)
  • molecules/proteins/atr.md — replication-stress kinase; ATR-Seckel mouse premature aging (Murga 2009; ~5 mo median lifespan from Fig 3a, NOT “~6 mo” overstatement); pancytopenia (NOT lymphopenia)
  • molecules/proteins/mrn-complex.md — MRE11-RAD50-NBS1 heterotrimer DSB sensor; Hopfner 2002 zinc-hook (150-600 Å arm range; C2G mutant 95-fold radiation sensitization); Lee & Paull 2005 ATM activation mechanism
  • interventions/pharmacological/dna-pkcs-inhibitors.md — class page; Park 2017 NU7441 in HFD food pellets (specific mg/kg dose NOT in paper); peposertib MTD NOT reached (RP2D=400 mg BID); SCID safety barrier as translation-blocked-safety

R32 verification statistics:

  • 15 pages verified end-to-end
  • ~80 source-text-level corrections
  • ~25 scientifically critical corrections, including:
    • Park 2017 NU7441 dose FABRICATED: “5 mg/kg i.p.” (DNA-PKcs-inhibitors page) and “40 mg/kg/day oral gavage” (dna-pkcs page) both wrong — paper uses HFD food pellets without stated mg/kg
    • Van Bussel 2020 multiple errors: MTD wasn’t reached (RP2D=400 mg BID, not “MTD 400 mg/day”); zero partial responses (was claimed 3/21); n=31 (was 21); median treatment 6.0 weeks (was 7.6)
    • Mouchiroud 2013 PARP inhibitor identity: PJ34 wasn’t used; actual inhibitors are AZD2281 + ABT-888
    • Heyn 2012 / Bocklandt 2011 patterns from R31b continue: many widely-cited “review-derived” numbers don’t match primary papers
    • Dang 2009 vs Das 2009 DOI confusion caught by histone-acetylation seeder
    • Pellegrini 2002 Nature substituted for “Yang 2002 Science” (BRC-RAD51 structure)
    • Lee & Paull 2005 ATM-activation timing “within seconds” → 30 min (Carney 1998 Nelms 1998 reference)
    • NBS phenotype list refined (removed unsourced “growth retardation” + “hypogammaglobulinemia”)
    • NU7441 vs NU7026 distinction: Veuger 2003 actually used NU7026 (correctly cited; my brief had it right by accident)
    • Bentley 1996 LIG1 lethality: “embryonic” → “perinatal hematopoietic-specific” (paper title explicitly says “not essential for mammalian cell viability”)
    • MUTYH active site UniProt-verified D131/D233 (546 aa isoform); my brief’s D138/E37 were not in any source
    • Al-Tassan 2002 KRAS attribution removed (paper has no KRAS data; only APC G:C→T:A signature)
    • Vermeij 2016 DR ramp protocol clarified (10% wk 7 → 30% wk 9, not direct 30%)
  • ~30 wrong DOIs/citations caught (Murga 2009 ng.351→ng.420; Lee & Paull 2005 1115297→1108297; Carney 1998 81524-5→81175-7; Stewart 1999 81560-9→81547-0; Jaiswal 2014 NEJMoa1409617→1408617; “Kane 1997” misattributed for Herman 1998; Wooster 2002 wrong; Mardalsa 2017 doesn’t exist; Sweasy 2022 substituted; many more)

Cross-cutting propagation done by main agent (R32 post-verification pass):

  • pathways/base-excision-repair.md — TWO critical corrections: (1) mtBER gap-filling polymerase: Polγ-exclusive → Polβ-major-via-Baptiste-2021 (POLβ outperforms POLγ ~3-fold in concentration, 30-fold in time-to-completion); (2) short-patch BER ligase: LIG1 → XRCC1-LIG3α (LIG1 is required for long-patch only — Levin 2000 / Pascucci 1999). New Baptiste 2021 footnote added.
  • molecules/proteins/parp1.md — Masson 1998 footnote DOI corrected (was wrong Mutat Res s0027-5107 → actual MCB 10.1128/mcb.18.6.3563); XRCC1 mechanism rewritten (BRCT1 direct protein-protein contact with auto-PARylated PARP1, not “generic PAR-chain reader” oversimplification)
  • molecules/proteins/dna-pkcs.md — Park 2017 NU7441 fabricated “40 mg/kg/day oral gavage” dose corrected to “HFD food pellets; specific mg/kg not stated in paper”; rhesus-specific basis of “~3-fold elevation in aged muscle” clarified (NOT generic; from rhesus n=5/group 15 yr vs 1 yr)

SOP-effectiveness check (R32):

  • New SOP § 3 (study-page-first) was used effectively across the batch:
    • MSH2 + MLH1 verifiers: Le 2015 NEJM cross-checked against verified [[mismatch-repair]] page — neither verifier re-read the PDF (estimated ~150K tokens saved)
    • PARP-inhibitors verifier: Bryant 2005 + Mouchiroud 2013 cross-checked from verified [[brca2]] and verified [[base-excision-repair]] pages
    • ATR verifier: Zou & Elledge 2003 (closed-access) cross-checked against verified [[rpa]] page — saved a re-read of an unavailable PDF
    • DNA-PKcs-inhibitors verifier: Pao 2020 + McIntyre 2019 + McDonald 2013 cross-checked from verified [[hdac]] page
    • MRN-complex verifier: Lee & Paull 2005 (closed-access) ATM-activation mechanism cross-checked against verified [[atm]] page
  • The pattern is now well-established. Estimated ~5-8 PDF re-reads avoided across R32; ~500K-1M input tokens saved.

Schema escalations:

  • mendelian-randomization study design (R31b carry-forward) still not in CLAUDE.md enum
  • translation-blocked-safety value used for DNA-PKcs-inhibitors (structural SCID barrier) — appropriate per existing schema, but a clarifying note in CLAUDE.md would help future seeders distinguish trial-bottleneck from safety-bottleneck
  • Heterotrimer protein page convention now well-established across [[ku70-ku80]], [[shelterin]], [[rpa]], [[mrn-complex]] — convention is stable
  • The seeder-brief DOI memory unreliability pattern continues across 4 consecutive rounds (R29-R32: ~80+ wrong DOIs caught total) — should be elevated to a structural CLAUDE.md note

Hallmark-level impact:

  • genomic-instability mechanistic depth before R32: 23 prot / 13 path / 2 proc / 7 interv (per 2026-05-07 lint table)
  • After R32: ~36 prot / 13 path / 2 proc / 9 interv. The intervention layer is now better-served (PARP inhibitors + DNA-PKcs inhibitors classes added). Mechanism layer near-complete for canonical DNA-repair pathways (BER + NER + HR + MMR + DDR-signaling all fully seeded).

Remaining gaps in this cluster (deferred):

  • TDG protein page (BER glycosylase; required for OSK rejuvenation per Lu 2020) — deferred from R31b
  • MSH3, MSH6, PMS2, EXO1 — MMR pathway completeness (msh2/mlh1 done)
  • POLG, POLG2, Twinkle — mtDNA replication (deferred from R30 inventory)
  • BRCA1 + BRCA2 study pages (Bryant 2005, Farmer 2005, Sharan 1997, Wooster 1995) — high-leverage given multi-page citation
  • Park 2017 + Niedernhofer 2006 + Vermeij 2016 study pages — high-leverage aging-context primary sources, would benefit from study-page-first SOP

Graph View

Backlinks