🧬 Aging Wiki

R3

50 min read

log/R3.md — Round 3 entries

Sub-file of log — see parent for index.

[2026-05-04] verify | cell-types/hematopoietic-stem-cells.md

Page verified: cell-types/hematopoietic-stem-cells.md

Sources checked:

  • Rossi 2007 (10.1038/nature05862) — local PDF verified
  • Beerman 2010 (10.1073/pnas.1000834107) — PMC PDF downloaded and verified
  • Sudo 2000 (10.1084/jem.192.9.1273) — PMC PDF downloaded and verified
  • Pang 2011 (10.1073/pnas.1116110108) — PMC PDF downloaded and verified
  • Jaiswal 2014 (10.1056/NEJMoa1408617) — bronze OA PDF downloaded and verified
  • Genovese 2014 (10.1056/NEJMoa1409405) — PDF blocked by Cloudflare CAPTCHA; PubMed abstract verified (n, gene list, HR confirmed sufficient)
  • Pinho & Frenette 2019 (10.1038/s41580-019-0103-9) — green OA but PMC retrieval failed; no-fulltext-access
  • Akunuru & Geiger 2016 (10.1016/j.molmed.2016.06.003) — closed-access (not_oa); no-fulltext-access

Corrections made:

  • Sudo 2000 footnote age cohorts (CORRECTION): wiki said “aged 2, 12, 24 months” — paper used 2, 6, 12, and 18 months. No 24-month cohort exists in Sudo 2000. Corrected footnote.
  • HSC expansion fold-change (CORRECTION): wiki said “~10× more immunophenotypic HSCs (LSK CD150+CD48−)” attributed to Sudo 2000. Sudo 2000 does NOT use SLAM markers (CD150/CD48) — it uses the CD34⁻KSL gate. The paper reports ~17-fold expansion (0.005% → 0.087%) from 2 to 18 months. “LSK CD150+CD48-” gating was introduced by Kiel et al. 2005. Corrected to “~17-fold” with explicit gate and source note. Tagged the “~10×” figure as unsourced.
  • Beerman 2010 myeloid bias mechanism (CORRECTION): wiki said “My-biased, balanced, and lymphoid-biased HSC subpopulations” — Beerman 2010 identifies only Slamf1^high (myeloid-biased) and Slamf1^low (balanced) subsets within Lin⁻Sca-1⁺c-Kit⁺flt3⁻CD34⁻. No “lymphoid-biased” HSC subpopulation is characterized in this paper. Corrected.
  • Beerman 2010 footnote n (CORRECTION): “n=not specified” changed to “n=3–10 recipients per group” with age specifications (young=3–4 mo, old=22–24 mo).
  • Pang 2011 “~3-fold” increase (CORRECTION): no specific fold-change number appears in the Pang 2011 text. The paper reports statistical significance (p<10⁻⁷) but fold-change must be read from Fig. 1B. Removed unsourced “~3-fold” and replaced with precise description of what the paper actually states (n, ages, p-value, gate used including CD45RA⁻).
  • Jaiswal 2014 prevalence threshold (CORRECTION): wiki said “~10% of individuals >65 years” — Jaiswal 2014’s discussion states “>70 years.” Age-stratified data: 60–69 = 5.6%, 70–79 = 9.5%, 80–89 = 11.7%, 90+ = 18.4%. Changed to “>70 years” for Jaiswal; “>65 years” is correctly attributed to Genovese 2014.
  • Jaiswal 2014 cardiovascular risk (CORRECTION): wiki said “~40% increased risk of cardiovascular events” — the paper reports HR=1.4 for all-cause mortality, and separate HR=2.0 for coronary heart disease and HR=2.6 for ischemic stroke. There is no composite “cardiovascular events” HR. Corrected to list actual reported HRs.
  • Genovese 2014 “WGS-based” (CORRECTION): wiki said “WGS-based detection” — Genovese 2014 used whole-exome sequencing (WES), not whole-genome sequencing. Corrected in text and footnote.
  • CHIP mutation characterization (CORRECTION): wiki said “most commonly in epigenetic regulators and splicing factors” — splicing factors (SF3B1=27 variants, SF3B1 not top genes) are substantially less frequent than DNMT3A (403 variants), TET2 (72), ASXL1 (62) in Jaiswal 2014. “And splicing factors” removed from “most commonly” phrasing; added as a separate qualified sentence.

Unverifiable claims (closed-access or failed download):

  • Pinho & Frenette 2019: niche description (endosteal vs perivascular distinction, CXCL4 involvement, aging niche changes) — all tagged no-fulltext-access.
  • Akunuru 2016: heterochronic transplant interpretation (“intrinsic changes are dominant”) — tagged no-fulltext-access.
  • CD49f as human HSC marker: not sourced to any paper verified in this pass (likely Notta et al. 2011, not cited on page).

Downstream propagation needed:

  • hallmarks/stem-cell-exhaustion.md — if it cites the “~10×” HSC expansion or attributes it to Sudo 2000 with SLAM markers, correction needed.
  • hallmarks/genomic-instability.md — if it cites Rossi 2007 HSC quiescence-DDR claim, verify framing is consistent.
  • studies/jaiswal-2014-chip-adverse-outcomes.md — if this study page exists, update n, HR values, and prevalence threshold to match verified values.
  • studies/genovese-2014-chip-cancer-risk.md — if this study page exists, correct “WGS-based” to “whole-exome sequencing.”

[2026-05-04] verify | cell-types/satellite-cells.md

Page verified: cell-types/satellite-cells.md Sources checked:

  • Mauro 1961 (10.1083/jcb.9.2.493) — bronze OA; PDF downloaded and verified
  • Conboy 2005 (10.1038/nature03260) — local PDF verified; corrections applied to strain description and footnote detail
  • Snijders 2015 (10.3389/fphys.2015.00283) — gold OA; PDF downloaded and verified
  • Conboy/Rando 2003 (10.1126/science.1087573) — closed-access (not_oa); Crossref abstract cross-checked; no-fulltext-access retained on footnote
  • Brack 2007 (10.1126/science.1144090) — closed-access (not_oa); Crossref abstract cross-checked; body claims consistent with abstract; no-fulltext-access retained on footnote
  • Sousa-Victor 2014 (10.1038/nature13013) — closed-access (not_oa); Crossref title/metadata + Snijders 2015 secondary citation cross-checked; body claims consistent; no-fulltext-access retained on footnote

Corrections made:

  • Conboy 2005 strain (CORRECTION): body text and footnote said “C57Bl/6 mice” for both young and old groups. Per the paper Methods, young partners were C57Bl/Ka-Ly5.2 or b-actin-eGFP transgenic (backcrossed ≥10 generations to C57BL/Ka-Thy1.1); only old mice were C57Bl/6. Corrected body text and footnote.
  • Conboy 2005 footnote enriched: added key quantitative detail — n=3–6 pairs per condition, P<0.005 for regeneration index and Delta upregulation, <0.1% GFP+ myotubes in aged muscle (confirming resident cell, not engraftment, mechanism).
  • Mauro 1961 footnote corrected: “model: frog and rat” tightened to “frog (tibialis anticus) primary; rat (sartorius) via personal communication from Palade” — the rat data was a personal communication, not the author’s own EM work. “Sarcolemma and basal lamina” noted; Mauro’s original text uses “plasma membrane” and “basement membrane.”
  • Archive status updated: Mauro 1961 and Snijders 2015 footnotes changed from “PDF pending download” to “PDF downloaded.”
  • Snijders 2015 footnote enriched: added full author list and note that c-Met is not recommended for human satellite cell enumeration and that NCAM/CD56 is preferred.
  • Marker table: CD34 human status changed from ”+” to ”+/-” with qualification note per Snijders 2015’s finding that c-Met/CD34 are less reliable than NCAM/CD56 for human cryosection enumeration.
  • Frontmatter key-markers-human: removed CD34+ (not well-supported per Snijders 2015 for human-specific identification).

Unverifiable claims (closed-access):

  • Conboy/Rando 2003: “3-fold reduction in Notch-activated cells post-injury” — cannot verify numerically; no-fulltext-access on footnote. Crossref abstract confirms general finding.
  • Brack 2007: mouse ages listed as “2 mo vs 21 mo” — cannot verify exact ages from abstract alone; no-fulltext-access.
  • Sousa-Victor 2014: geriatric age range “28–32 mo,” specific age groups “3 mo young vs 20 mo old vs 28–32 mo geriatric,” p16 silencing transplant rescue details — cannot verify without full PDF; no-fulltext-access.

Downstream pages checked:

  • sarcopenia.md: cites Conboy 2005 via [^conboy-2005] footnote. Claims are accurate and consistent with verified Conboy 2005 findings. No correction needed.

[2026-05-04] verify | model-organisms/heterocephalus-glaber.md

Page verified: model-organisms/heterocephalus-glaber.md Sources checked:

  • Ruby 2018 (10.7554/elife.31157) — local PDF verified (gold OA)
  • Ruby 2023 (10.7554/elife.88057) — local PDF verified (gold OA)
  • Tian 2013 (10.1038/nature12234) — local PDF verified (green OA via PMC)
  • Takasugi 2021 / del Marmol 2021 (10.1038/s41598-021-86967-9) — local PDF verified (gold OA)
  • Seluanov 2009 (10.1073/pnas.0905252106) — local PDF verified (green OA)
  • Tian 2015 (10.1073/pnas.1418203112) — local PDF verified (bronze OA)
  • Kim 2011 (10.1038/nature10533) — local PDF verified (hybrid OA via PMC)
  • Park 2017 (10.1126/science.aab3896) — DOI lookup failed; Cambridge OA URL too large for WebFetch; tagged no-fulltext-access for quantitative anoxia survival claims
  • Edrey/Buffenstein 2011 (10.1016/j.exger.2010.09.005) — closed-access (not_oa); not verified
  • Hadj-Moussa 2021 (10.1002/jcp.30216) — closed-access (not_oa); not verified

Corrections made:

  • Ruby 2018 caste disaggregation (MAJOR): wiki stated dataset “does not disaggregate by caste” — WRONG. Figure 3 of Ruby 2018 explicitly disaggregates by sex × breeding status (4 sub-populations); all show non-increasing hazard; breeders have ~5–10× lower hazard than non-breeders. Corrected.
  • Ruby 2023 n: “n=~6,000+” → “n=6,893 qualifying animals (from 7,536 catalogued).” Added eLife assessment note (evidence rated incomplete/inadequate for older ages).
  • Seluanov 2009 “~75% confluence” removed: unsourced — paper never states this figure. Replaced with paper’s actual language (cell density >3× lower than mouse).
  • Seluanov 2009 p16/p27 direction corrected: wiki inverted the mechanism. p16^INK4a mediates ECI (early, low-density); p27^Kip1 mediates regular contact inhibition (high density). Both layers present in NMR, but p16 is the novel early layer, not p27.
  • Takasugi 2021 characterization (MAJOR): wiki said paper “confirmed excess HA abundance” — INCORRECT. Paper disputes 6–12 MDa figure; finds NMR HA max ~2.5 MDa; no ultra-HMW (≥3 MDa) detected; NMR fibroblasts do not secrete ultra-HMW HA in vitro; Alcian Blue (Tian 2013 method) found non-HA-specific. HA abundance higher in NMR confirmed (most tissues), MW systematically higher than GP but not ultra-HMW. Added full contradiction description + contradictory-evidence.
  • Park 2017 anoxia claim tagged needs-replication: PDF unverifiable.
  • Removed auto-extraction banner.

Unverifiable claims:

  • Park 2017 ~18 min anoxia survival: DOI lookup failed; PDF inaccessible.
  • Edrey/Buffenstein 2011 (review): closed-access; eusocial biology, IGF1, pain biology, proteasome claims not independently verified.
  • Hadj-Moussa 2021: closed-access; microRNA hypoxia claim not verified.
  • AnAge 31-year max lifespan, NCBI Taxonomy 10181, genome 2,670 Mb: canonical database fields not re-checked against live databases.

Downstream propagation needed:

  • hypotheses/negligible-senescence.md — may not reflect caste-disaggregated Ruby 2018 data or Ruby 2023 n correction.
  • Any page citing Takasugi 2021 as “confirming HMW-HA” needs correction — paper disputes the claim.
  • cancer-resistance page (if exists) — HMW-HA section may inherit Tian 2013 6–12 MDa figure without the Takasugi 2021 dispute.

[2026-05-04] verify | model-organisms/nothobranchius-furzeri.md

Page verified: model-organisms/nothobranchius-furzeri.md Sources checked:

  • Valenzano 2006 (10.1016/j.cub.2005.12.038) — PDF downloaded and verified (bronze OA via camoufox)
  • Reichwald 2015 (10.1016/j.cell.2015.10.071) — PDF downloaded and verified (bronze OA via camoufox)
  • Smith 2017 (10.7554/eLife.27014) — PDF downloaded and verified (gold OA)
  • Terzibasi 2009 (10.1111/j.1474-9726.2009.00455.x) — PDF downloaded and verified (bronze OA)
  • Cellerino 2016 (10.1111/brv.12183) — PDF downloaded and verified (hybrid OA)
  • Dolfi 2019 (10.1186/s13227-019-0142-5) — PDF downloaded and verified (gold OA)
  • Graf 2013 (10.1016/j.exger.2012.02.012) — not_oa; closed-access; unverified; no-fulltext-access added to telomere in-vitro claims
  • de Bakker & Valenzano 2023 (10.1016/j.arr.2023.102019) — PDF download failed (bronze OA; repository link failed); unverified; no-fulltext-access added to brain aging section

Corrections made:

  • GRZ strain lifespan (MAJOR): frontmatter typical-lifespan corrected from “GRZ lab strain ~4 months median” → “GRZ median ~9 weeks (~2 months); maximum ~13 weeks (~3 months)” per Terzibasi 2009 Table 1 and Valenzano 2006 Discussion (captive median 9 weeks). Cellerino 2016 review reports “as short as 3 months.” The “4 months” figure was inflated by ~2×.
  • Identity table Typical median lifespan: corrected from “4 months (GRZ); 6–9 months (wild-derived)” to actual strain-specific values from Terzibasi 2009 Table 1.
  • Strain table (MZCS-0803 / wild-derived row): replaced vague “~8–9 months” estimates with Terzibasi 2009 Table 1 actuals: MZM-0410 mean 20 wk/max 26 wk; MZM-0403 mean 24 wk/max 32 wk.
  • Genome size: frontmatter genome-size-mb corrected from ~1200 to ~1240 (Reichwald 2015 Table 1: final assembly = 1,242,498,532 bp); identity table similarly corrected.
  • Genome assembly name: corrected from unsourced “Nfu_20140520” to actual GenBank accessions KG817100/KG959958 per Reichwald 2015.
  • genome-similarity-to-human frontmatter field: removed fabricated “~70% one-to-one vertebrate orthologs” claim (NOT in Reichwald 2015); replaced with sourced CEGMA 98% coverage figure plus unsourced note.
  • Valenzano 2006 resveratrol section: added dose-specific quantitative effects (+33%/+27% median/max at 120 µg/g; +56%/+59% at 600 µg/g; lowest dose 24 µg/g ineffective), test statistic (log-rank p<0.001), n’s (n=157 total), and clarified mechanism uncertainty.
  • Valenzano 2006 footnote: upgraded from stub (no n, no effect size) to full citation with n=157, effect sizes, p-value.
  • Smith 2017 FMT section: added quantitative lifespan effects (+37% vs wt p=4.04×10⁻⁹; +41% vs same-age controls p=5.08×10⁻⁶; +21% vs antibiotic-only p=5.89×10⁻⁵), clarified four-arm experimental design, added locomotor activity finding, corrected recipient age to 9.5 weeks and donor age to 6 weeks.
  • Smith 2017 footnote: upgraded from stub to full citation with group design, p-value, effect sizes.
  • Terzibasi 2009 DR section: added strain-specific effect sizes (GRZ: median +13%, p=0.005; MZM-04/10P: max extended p=0.02, no median effect), specified n=10/group for behavior, noted early mortality in wild-derived strain.
  • Terzibasi 2009 footnote: replaced placeholder “n=reported per group” with actual n=10/group (behavior) and actual strain lifespans.
  • Interventions summary table: quantified all three interventions with actual effect sizes.
  • Diapause section: corrected “G1-like quiescent state” → “specifically G1 phase” (confirmed by FUCCI reporter experiments in Dolfi 2019; diapause I arrest is in G1, exit is synchronous into S phase directly).
  • Comparison table lifespan row: corrected from “4–9 months” to “2–8 months (strain-dependent)”.
  • Intro paragraph: corrected “~4–6 months in standard lab strains” to “~9 weeks (~2 months) in GRZ inbred” with supporting citations.
  • Brain aging section: tagged with no-fulltext-access for de Bakker 2023 claims.
  • Graf 2013 footnote: tagged UNVERIFIED + closed-access + no-fulltext-access.

Pages with potentially stale downstream claims:

  • molecules/compounds/ — any fisetin or resveratrol page citing the killifish resveratrol result should note the corrected GRZ strain lifespan baseline
  • interventions/pharmacological/senolytics.md or any page citing Smith 2017 microbiome extension — may need the quantitative +37% figure added
  • Any page citing killifish “4–6 month” lifespan as the species baseline — should be corrected to GRZ-specific (~9 wk) vs wild-derived (5–6 mo)

[2026-05-04] verify | model-organisms/saccharomyces-cerevisiae.md

Page verified: model-organisms/saccharomyces-cerevisiae.md Sources checked:

  • Kaeberlein 1999 (10.1101/gad.13.19.2570) — local PDF verified (diamond OA)
  • Kaeberlein 2004 (10.1371/journal.pbio.0020296) — downloaded and verified (gold OA; PMC)
  • Heitman 1991 (10.1126/science.1715094) — not_oa; verified via Crossref abstract only
  • Kaeberlein 2005 (10.1126/science.1115535) — not_oa; verified via Crossref abstract only
  • Fabrizio 2001 (10.1126/science.1059497) — not_oa; verified via Crossref abstract only
  • Tsukada/Ohsumi 1993 (10.1016/0014-5793(93)80398-e) — download failed (no PMC record); verified via Crossref abstract only; no-fulltext-access retained

Corrections made:

  • CR/SIR2-independent claim (MAJOR): wiki stated “CR extends RLS even in sir2Δ cells” — WRONG per Kaeberlein 2004 Figure 2A. CR alone does NOT extend RLS in sir2Δ single mutants. The SIR2-independent effect is demonstrated only in sir2Δ fob1Δ double-mutant background where ERC levels are already reduced. Body text corrected to accurately describe the experimental context.
  • Kaeberlein 2004 footnote: updated to specify BY4742 WT mean RLS of 27 generations (Table 1); specified n=20 mother cells per strain; corrected CR effect size description; noted strain-background dependency (PSY316 unresponsive); updated download status to locally available.
  • Kaeberlein 1999 footnote: strain corrected from “W303” to “W303R” (W303-1A MATa RDN1::ADE2 RAD5); added actual mean RLS values (sir2Δ 11.6 vs W303R 21.4; SIR2-OE 28.8 vs W303R 23.0 generations, n=50).
  • Kaeberlein 2005 body text: removed unverified “~60% RLS extension” claim (not stated in Crossref abstract; source is closed-access); replaced with verified description from abstract (TOR/Sch9 epistatic to CR; Sir2-independent).
  • Fabrizio 2001 body text and footnote: corrected from “adenylate cyclase” attribution; flagged RAS2-specific deletion claim as unverifiable from abstract; added 3-fold CLS extension figure per abstract; added no-fulltext-access.
  • Ohsumi 1993 footnote: corrected gene count from “apg1–apg10” (10 genes implied) to “15 complementation groups” per Crossref abstract; added download failure status and no-fulltext-access.
  • Heitman 1991 footnote: added verified title; expanded mechanism description per Crossref abstract; flagged as abstract-only verification.

Unverifiable claims:

  • Kaeberlein 2005 “~60% RLS extension” — not stated in abstract; source closed-access. Removed from body, flagged in footnote.
  • Fabrizio 2001 RAS2 deletion — abstract mentions adenylate cyclase (CYR1) mutations and SCH9, not RAS2 specifically. Cannot verify from abstract.
  • Tsukada/Ohsumi 1993 quantitative details — abstract-only; download failed.

Downstream propagation needed:

  • molecules/proteins/sirt1.md — cites the same Kaeberlein 1999 footnote; numbers there (23.0 vs 28.8, n=48-50) are consistent with Figure 7A of the paper and already correctly stated.
  • processes/caloric-restriction.md or interventions/lifestyle/caloric-restriction.md — may cite the Kaeberlein 2004 CR/SIR2-independent result; check for the erroneous “sir2Δ single-mutant” framing.

Final verified state: verified: true with scope (Kaeberlein 1999 and 2004 PDFs fully verified; other sources abstract-only or closed-access).

[2026-05-04] verify | model-organisms/rattus-norvegicus.md

Page verified: model-organisms/rattus-norvegicus.md Sources checked:

  • Rat Genome Sequencing Project Consortium 2004 (10.1038/nature02426) — local PDF verified
  • Masoro, Yu & Bertrand 1982 PNAS (10.1073/pnas.79.13.4239) — downloaded and verified
  • Anisimov 2010 (10.18632/aging.100230) — downloaded and verified
  • McCay 1935 (10.1093/jn/10.1.63) — closed-access; unverifiable; no-fulltext-access retained
  • Yu/Masoro 1985 (10.1093/geronj/40.6.657) — closed-access; unverifiable; no-fulltext-access retained
  • Moran 2005 (10.1007/s10522-005-4808-0) — closed-access; unverifiable
  • Burwell 1994 (10.1016/0197-4580(94)90027-2) — download failed; unverifiable
  • Gallagher 2002 (10.1002/hipo.10016) — closed-access; unverifiable

Corrections made:

  • Masoro 1982 characterization: wiki described it as a “review” that “summarized mechanistic evidence that CR delays biological aging processes” — WRONG. It is a primary data paper (n=115/group, Fischer 344 male rats, 60% CR) arguing AGAINST the metabolic rate hypothesis. Corrected body text and footnote to reflect actual paper thesis and data (mean LS: 701d ad lib vs 986d restricted, ~41% extension).
  • Anisimov 2010 rat strains: wiki claimed “female outbred LIO and CBA rat strains” — CBA does not appear in Anisimov 2010 at all. CBA is absent from Table 1. Rat strains in this paper are LIO (female) and Fischer-344 (male). Corrected.
  • SHR organism confirmed mice: Anisimov 2010 Table 1 lists SHR under Mouse section; ref [70] = “Metformin slows down aging and extends life span of female SHR mice” (Cell Cycle 2008). Confirmed SHR = mice, not rats. Body text clarification updated with primary-source citation.
  • Interventions table: metformin rat entry changed from “LIO, CBA” to “female outbred LIO and male Fischer-344”; description corrected to reflect actual results (buformin modest effect in LIO; metformin no significant effect in F344).
  • Anisimov footnote: removed “CBA outbred strains”; added clarification that SHR data are mice; marked PDF verified.
  • Removed ⚠️ auto-extraction banner.

Unverifiable claims (closed-access sources):

  • McCay 1935 effect sizes for albino rats — exact quantitative figures cannot be verified
  • Yu/Masoro 1985 “30–40% longer” lifespan figure for F344 CR rats — cannot verify against primary source
  • Moran 2005 sarcopenia model claims (F344 vs F344×BN comparisons) — cannot verify
  • Burwell 1994 behavioral aging battery results (F344 vs BN vs F1) — cannot verify
  • Gallagher 2002 hippocampal PKCγ / MWM correlation claims — cannot verify

Downstream pages that may need updates:

  • molecules/compounds/metformin.md (or equivalent) — cited Anisimov SHR as rats; now confirmed mice; may carry the “CBA” error
  • interventions/lifestyle/caloric-restriction.md — cites Masoro 1982 possibly as a review; may need same correction
  • Any page citing Anisimov 2010 rat data should be checked for CBA strain reference

[2026-05-04] verify | model-organisms/homo-sapiens.md

Page verified: model-organisms/homo-sapiens.md Sources checked:

  • Willcox 2008 (10.1073/pnas.0801030105) — local PDF verified
  • Suh 2008 (10.1073/pnas.0705467105) — local PDF verified
  • Lopez-Otin 2013 (10.1016/j.cell.2013.05.039) — pre-verified; not re-read
  • Horvath 2013 (10.1186/gb-2013-14-10-r115) — downloaded and verified
  • Waziry 2023 (10.1038/s43587-022-00357-y) — downloaded and verified
  • Ravussin 2015 (10.1093/gerona/glv057) — downloaded and verified
  • Sebastiani 2012 (10.3389/fgene.2012.00277) — downloaded and verified
  • Kraus 2019 (10.1016/s2213-8587(19)30151-2) — download failed; tagged no-fulltext-access
  • Coles 2004 (10.1093/gerona/59.6.b579) — download failed; tagged no-fulltext-access

Corrections made:

  • Willcox 2008 footnote n: “n=3,741 Japanese-American men” → “n=615 nested case-control (213 cases + 402 controls; drawn from 3,741-person HHP/HAAS cohort)”; added exact 95% CI (1.51–5.02) and p=0.0007; design corrected to “nested case-control”
  • Suh 2008 body text: “carriers had ~35% lower serum IGF-1 activity” → corrected to: 9/384 centenarians (2.3%) carried IGF1R nonsynonymous mutations vs 1/312 controls (0.3%), p=0.02; carriers showed reduced IGFIR signaling activity; female offspring of centenarians (not mutation carriers) had 35% higher serum IGF-1 — a compensatory response to reduced receptor sensitivity; added 2.5 cm shorter stature finding (p<0.001)
  • Horvath 2013 footnote: “n=~8,000 samples” → “n=7,844 non-cancer samples from 82 datasets (51 tissues/cell types)”; added “test-set MAE=3.6 yr” to footnote
  • Waziry 2023 body text (critical): “CALERIE showed 2-year 25% CR reduced epigenetic age acceleration by ~2.5 years” → corrected to: CR achieved only ~11.9% (not 25%); DunedinPACE (pace of aging) slowed by ~2–3% per year (d=−0.25 at 24 months, 95% CI −0.41 to −0.09, p<0.003); PhenoAge and GrimAge were NOT significantly changed; “2.5 year” figure is not in the paper
  • Waziry 2023 footnote: “n=196 (subset)” → “n=197 with baseline + follow-up DNAm data (128 CR, 69 AL)”
  • CALERIE cohort table: “218” → “220 randomized (218 started)”; added “~11.7% achieved”
  • Ravussin 2015 divergent systems table: “no mortality benefit at 2 yr and only modest biomarker improvement” → corrected to: achieved 11.7% CR (not 25%); biomarker improvements were significant (TNF-α, CRP, T3, lipids, RMR); mortality was not an endpoint in this 2-year trial in 21–51 year olds; added no-fulltext-access for Kraus 2019
  • Ravussin 2015 footnote: “n=218” → “n=220 randomized (218 ITT; 145 CR, 75 AL)”; added “achieved 11.7±0.7% CR”
  • Sebastiani 2012 footnote: “n=~800 centenarians + controls · observational” → “review (NECS overview 1994–2012; GWAS sub-study: 801 centenarians + 914 controls)”
  • Coles 2004 footnote: added no-fulltext-access tag
  • Kraus 2019 footnote: added no-fulltext-access tag
  • Banner removed; verified flag flipped to true (partial scope: Kraus 2019 + Coles 2004 unverifiable)

Unverifiable claims:

  • Max-verified lifespan 122.5 yr (Calment) cited to Coles 2004 — well-established historical fact; Coles 2004 PDF inaccessible but claim is independently corroborated
  • Kraus 2019 CALERIE cardiometabolic claims — PDF inaccessible; claims tagged no-fulltext-access

Downstream propagation needed: Main agent should check:

  • studies/willcox-2008-foxo3a-longevity — update n and design if extracted from this page
  • studies/suh-2008-igf1r-centenarian — update IGF-1/IGFIR mechanism description
  • studies/waziry-2023-calerie-epigenetic-clock — if it says “2.5 yr slowing” anywhere, correct to DunedinPACE d=−0.25 / ~2–3% slowing of pace
  • interventions/lifestyle/caloric-restriction.md — may inherit wrong “2.5 yr” figure or “25% CR achieved”
  • molecules/proteins/foxo3.md — may inherit wrong Willcox 2008 n

[2026-05-04] verify | model-organisms/mus-musculus.md

Page verified: model-organisms/mus-musculus.md Sources checked: All 7 cited DOIs verified against local PDFs.

Corrections made:

  • Ortholog %: “~85% one-to-one orthologs” → “~80%” (MGSC 2002 explicitly states 80% reciprocal best-match 1:1; ~99% have any human homologue)
  • Harrison 2009 rapamycin dose: “600 ppm” → “14 mg/kg food” (paper Methods: “administered at 14 mg per kg food”)
  • Harrison 2009 effect sizes: “~10–14% both sexes” (median LS, unqualified) → “+9% ♂, +13% ♀ mean LS; +9% ♂, +14% ♀ at 90th percentile” (paper reports mean and 90th percentile, not median)
  • Holzenberger 2003 ITP table note: clarified the 26% is mean lifespan, both sexes combined — not females specifically; female result is +33%, male +15.9% (NS)
  • Interventions table: Holzenberger “+26% median LS (129/Sv ♀)” → “+26% mean LS combined; ♀ +33%, ♂ +15.9% NS”
  • Martin-Montalvo 2013: “+5–6% median LS” → “+5.83% mean LS” (paper uses Gehan-Breslow test on mean, not median)
  • Strong 2016 acarbose: “+22% ♂, ~5% ♀” → “+6% ♂ median, +2% ♀ (NS)” (Table 1, C2012 cohort, 1000 ppm started 16 months)
  • Strong 2016 17α-estradiol: “+12% ♂” → “+19% ♂” (Table 1, C2011 cohort, 14.4 ppm)
  • Prowse & Greider 1995: “constitutive telomerase in most somatic tissues” → “detected in liver, kidney, spleen, testis; absent from brain” (more precise per paper results)
  • Tabula Muris Senis footnote: corrected — DOI 10.1038/s41586-020-2499-y is Schaum et al. 2020 (organ bulk RNA-seq + TMS scRNA companion paper), not the main TMS scRNA atlas; noted correct TMS atlas DOI (10.1038/s41586-020-2496-1)
  • Banner removed; verified flag flipped to true

[2026-05-04] verify | molecules/proteins/bid.md

Page verified: molecules/proteins/bid.md

Sources checked (6 verified / 1 unverifiable):

  • Wang 1996 (10.1101/gad.10.22.2859) — local PDF, verified end-to-end (was pending at session start; downloaded via archive before read)
  • Li 1998 (10.1016/s0092-8674(00)81590-1) — local PDF, verified end-to-end
  • Luo 1998 (10.1016/s0092-8674(00)81589-5) — local PDF, verified end-to-end
  • Zha 2000 (10.1126/science.290.5497.1761) — local PDF, verified end-to-end
  • Chipuk 2010 (10.1016/j.molcel.2010.01.025) — local PDF, verified end-to-end
  • Kuwana 2002 (10.1016/s0092-8674(02)01036-x) — local PDF, verified end-to-end
  • Yin 1999 (10.1038/23730) — not_oa; confirmed DOI correct via Crossref (10.1038/22320 is wrong — “DOI not found”); BID KO phenotype claims unverified; tagged no-fulltext-access

Key corrections:

  1. CRITICAL — Caspase-8/granzyme B site species labels inverted throughout: wiki wrote “Asp59 (human) / Asp75 (mouse)” for caspase-8 cleavage. Both Li 1998 and Luo 1998 use murine BID and identify Asp59 (after LQTD motif, conserved in human) as the caspase-8 site. Asp75 (mouse, after IEPDS motif) is the granzyme B site. Labels were inverted. Corrected in lead paragraph, domain table, activation mechanism steps, cleavage-site note, and protease table.

  2. Kuwana 2002 — two fabricated claims removed: (a) “anti-cardiolipin antibodies blocked tBID’s membrane association” — the paper demonstrates cardiolipin requirement by liposome lipid composition comparison, with no antibody-blocking experiment; (b) “appeared as toroidal pores” — Fig 4 of Kuwana 2002 explicitly shows NO visible membrane disruptions by negative-stain and thin-section EM; the authors use the term “supramolecular openings.” Both claims replaced with accurate descriptions.

  3. Zha 2000 model corrected: “murine fibroblasts” → Jurkat cells (metabolic labeling with [³H]myristic acid) and MCF7/Fas cells (in vivo GFP redistribution); in vitro assays used isolated mouse liver mitochondria. Quantitative mitochondrial association data added: ~100% myr-p15 vs <30% G60A-p15 pelleted with sedimented mitochondria; amide linkage of myristate confirmed.

  4. Wang 1996 “8 α-helices” qualified: Wang 1996 did not characterize helical structure; the 8-helix count comes from subsequent NMR work (Chou et al. 1999, Cell 96:615). Note added to body text.

  5. Wang 1996 actual data added to footnote: murine T-cell hybridoma cDNA library, FL5.12 cells, 195 aa, predicted MW 21.95 kDa; BID lacks BH1/BH2/BH4/C-terminal anchor.

  6. Kuwana 2002 system corrected in footnote: “isolated outer membrane vesicles + liposomes reconstituted from Xenopus egg mitochondrial lipids”; fluorescein-dextran (10 kDa and 2 MDa) used as permeability markers; EM showed no visible pore morphology.

Unverifiable claims (1 source):

  • Yin 1999 — KO mouse phenotype (FasL resistance of Bid−/− hepatocytes, normal lymphopoiesis, enhanced Bcl-2 expression) — all tagged no-fulltext-access; DOI confirmed correct but paper is not_oa.

Final verified state: verified: true (partial scope — Yin 1999 not_oa; BH3 affinity Kd values and cathepsin/calpain cleavage claims unsourced; canonical-database identity fields not re-checked).

Downstream propagation needed:

  • Any page citing Li 1998 or Luo 1998 with “Asp59 (human) / Asp75 (mouse)” caspase-8/granzyme-B site labels — species attribution is inverted; correct to Asp59 = caspase-8 site (murine, conserved human); Asp75 = granzyme B site (murine)
  • Any page citing Kuwana 2002 with “toroidal pore” morphology or “anti-cardiolipin antibody blocking” — both claims are not from that paper; remove/correct
  • pathways/apoptosis-pathway.md — likely inherits cleavage-site labeling from bid.md
  • pathways/bcl-2-family-signaling.md — may describe tBID cleavage-site numbering; check for species inversion

Unverifiable claims: None — all 7 sources successfully downloaded and read.

Downstream propagation needed: Main agent should check:

  • molecules/compounds/rapamycin.md — may inherit “600 ppm” dose from Harrison 2009 reference
  • pathways/mtor.md — may cite Harrison 2009 effect sizes
  • molecules/compounds/metformin.md — may say “median” not “mean” for Martin-Montalvo result
  • molecules/compounds/fisetin.md — Tabula Muris Senis reference may use wrong DOI

Append-only record of significant ingests, lint passes, and structural changes. See CLAUDE.md § Logging for format.

[2026-05-04] verify | pathways/pi3k-akt-pathway.md

Page verified: pathways/pi3k-akt-pathway.md Sources checked: Holzenberger 2003 (local PDF), Willcox 2008 (local PDF), Engelman 2006 (local PDF), Manning & Toker 2017 (downloaded during pass), Vasan & Cantley 2022 (PMC abstract/full text only). WikiPathways WP4172 confirmed via direct HTTP.

Corrections made:

  • Holzenberger 2003 strain: “C57BL/6 background” → “129/Sv background” (PDF Methods explicitly states 129/Sv)
  • Holzenberger 2003 lifespan: table row relabelled from “females only” to both sexes; exact values confirmed: overall +26% (P<0.02, Cox), females +33% (756±46 vs 568±49 days, P<0.001), males +15.9% (679±80 vs 585±69 days, NS)
  • Citation key renamed from [^holzenberger2002] to [^holzenberger2003] (consistent with print year and verified insulin-igf1 page)
  • FOXO3A phosphorylation site ordering corrected: Ser253→Thr32→Ser315 reordered to T32, S253, S315 (N- to C-terminal per Manning 2017 Fig 4A)
  • Willcox 2008 footnote expanded with exact OR=2.75 (95% CI 1.51–5.02, P=0.0007), heterozygous OR=1.91 (95% CI 1.34–2.72, P=0.0003), exact n=615 (213 cases, 402 controls)
  • GH receptor row: “~30–40% lifespan extension” → “~25–55% lifespan extension (range across studies)” with needs-replication tag (uncited row)
  • Manning 2017 and Vasan 2022 footnotes updated to reflect download/verification status
  • ⚠️ banner removed; verified flag flipped to true

Downstream propagation needed: Main agent should check molecules/compounds/quercetin.md, interventions/pharmacological/senolytics.md, and any study pages that cite Holzenberger 2002 for the corrected strain (129/Sv).

[2026-05-04] verify | molecules/proteins/rictor.md

Page verified: molecules/proteins/rictor.md Sources checked:

  • Sarbassov 2004 (10.1016/j.cub.2004.06.054) — local PDF verified
  • Zinzalla 2011 (10.1016/j.cell.2011.02.014) — local PDF verified
  • Soukas 2009 (10.1101/gad.1775409) — local PDF verified
  • Jacinto 2006 (10.1016/j.cell.2006.08.033) — local PDF verified
  • Bentzinger 2008 (10.1016/j.cmet.2008.10.002) — local PDF verified (pre-downloaded)
  • Lamming 2012 (10.1126/science.1215135) — Crossref abstract only; MIT DSpace PDF blocked (403)
  • Sarbassov 2005 (10.1126/science.1106148) — closed-access (not_oa); not verifiable
  • Jacinto 2004 (10.1038/ncb1183) — closed-access (not_oa); not verifiable
  • Cybulski/Hagiwara 2012 (10.1016/j.cmet.2012.03.015) — Crossref abstract returned no abstract text; PDF pending
  • Mizunuma 2014 (10.1111/acel.12248) — PubMed abstract verified; PDF still downloading

Corrections made:

  • Rictor MW: “~190 kDa” → “192 kDa” (Sarbassov 2004 paper explicitly states 192 kDa)
  • Ribosome-binding domain: removed unsourced residue range “~521–570” — Zinzalla 2011 demonstrates ribosome association biochemically but defines no specific residue boundaries; relabelled “domain” → “region” with needs-replication tag
  • Soukas 2009 lifespan claim: INVERTED — wiki stated “extended lifespan on rich nutrient media”; source shows rict-1 mutants are SHORT-LIVED on standard OP50 diet (−24 to −43%); HB101 extension (~74%) is diet-mediated via reduced food consumption, not direct pro-longevity signaling from RICT-1 loss
  • Jacinto 2006 footnote authors corrected: “Frias M, Bhatt D” → actual authors Facchinetti V, Liu D, Soto N, Wei S, Jung SY, Huang Q, Qin J, Su B; footnote updated to reflect key finding (sin1 deletion abolishes Ser473 but not Thr308)
  • Sarbassov 2004 footnote: cell line corrected from “HEK293 cells” → “HEK293T cells, HeLa cells, DU145 cells”; author list completed; download status updated to locally downloaded
  • Bentzinger 2008 body text + footnote: added key finding that AKT Ser473 is still hyperphosphorylated in double-KO (raptor + rictor) muscles — mTORC2 NOT required for AKT Ser473 in skeletal muscle in vivo (tissue-specific exception to general model); survival data added (RAmKO ~110d; RImKO >2 years)
  • First-author correction: “Cybulski et al. 2012” → “Hagiwara et al. 2012” in body text; footnote corrected with full author list (Hagiwara A is first author)
  • Lamming 2012 footnote: full author list added; rapamycin dose flagged as unconfirmed from full text (#gap/dose-response-unclear)
  • Mizunuma 2014 footnote: PMID 25040785 added; abstract-level detail confirmed
  • Auto-extraction banner removed; verified flag flipped to true (partial scope documented in verified-scope field)

Unverifiable claims:

  • Sarbassov 2005 (closed-access): quantitative RNAi knockdown details and in vitro kinase assay specifics
  • Jacinto 2004 (closed-access): PKCα substrate identification quantitative details
  • Lamming 2012 rapamycin dose and n per group — full text inaccessible; MIT DSpace 403
  • All PTM sites in frontmatter and table — not cross-checked against primary MS proteomics papers

Downstream propagation needed: Main agent should check:

  • pathways/mtor.md — may cite rict-1 lifespan claim in inverted form
  • molecules/compounds/rapamycin.md — Lamming 2012 likely cited for mTORC2 insulin-resistance mechanism; dose unconfirmed
  • Any page citing “Cybulski et al. 2012” — first-author attribution needs correction to Hagiwara et al.

[2026-05-04] verify | molecules/proteins/sirt1.md

Page verified: molecules/proteins/sirt1.md Sources checked: Rodgers 2005 (local PDF, verified); Pacholec 2010 (local PDF, verified); Hubbard 2013 (local PDF, verified); Yeung 2004 (downloaded bronze OA, verified); Luo 2001 (downloaded bronze OA, verified); Satoh 2013 (downloaded bronze OA, verified); Kaeberlein 1999 (downloaded diamond OA, verified). Brunet 2004: not_oa, unverifiable.

Corrections made:

  • PGC-1α “Lys778 (and additional sites)” framing: Rodgers 2005 maps 13 lysine sites by MS but does not identify “Lys778” by name; Lys778 designation comes from Hubbard 2013 peptide substrate. Corrected in body and footnote.
  • Rodgers 2005 footnote: expanded with quantitative detail (SIRT1 ~1.7-fold increase fasted liver; liver NAD+ +33%; PGC-1α deacetylation ~60% with SIRT1+NAD+; model = 4-week-old C57BL/6 fasted 24h; specific gluconeogenic but not mitochondrial gene regulation).
  • AMPK Thr177/Ser538 sites on PGC-1α: re-tagged unsourced — these are from Jager et al. 2007 not Rodgers 2005.
  • N-terminal domain: functional STAC-binding sub-region corrected to residues ~190–244 (per Hubbard 2013 HDXMS), not residues 1–235 as a whole.
  • Satoh 2013 body temperature direction: “reduced body temperature” → “elevated rectal body temperature” (factual reversal; Fig. 2B shows higher Tg temperature).
  • Satoh 2013 lifespan numbers added: females ~16% (799→930 d, p<0.001), males ~9% (849→926 d, p=0.015), combined ~11% (835→926 d, p<0.001).
  • Satoh 2013 mechanism: specified Nkx2-1 K161/K182 deacetylation sites (from Fig. 4H).
  • Yeung 2004 cell model: “HEK293 and MEF cells” → NSCLC lines (H358, H460, H1299) as primary model; HEK293T and MEFs for specific assays.
  • Luo 2001 footnote: added apoptosis data (16.4% vs 32.3% SubG1), nicotinamide dose (5 mM), Sir2α/hSIRT1 identity note.
  • Kaeberlein 1999 footnote: added exact lifespan numbers (28.8 vs 23.0 generations for SIR2 OE; 11.6 for sir2Δ; n=48–50).
  • no-fulltext-access removed from Satoh 2013, Yeung 2004, Luo 2001, Kaeberlein 1999 (all now verified).
  • Limitations section updated; Brunet 2004 remains no-fulltext-access.
  • Banner removed; verified: true (partial scope — canonical-DB fields not re-verified).

Downstream propagation needed:

  • pathways/sirtuin.md — cross-check E230 allosteric site framing for consistency (N-terminal domain 190–244 vs 1–235).
  • Study pages seeded from Satoh 2013 — body temperature direction error likely propagated.
  • Any page citing SIRT1/FOXO3 deacetylation from Brunet 2004 — still rests on unverified closed-access source.

[2026-05-04] verify | molecules/proteins/s6k1.md

Page verified: molecules/proteins/s6k1.md Sources checked:

  • 10.1016/j.cmet.2006.05.003 (Um 2006) — local PDF downloaded and verified in full (10 pages)
  • 10.1016/j.cell.2005.10.024 (Holz 2005) — local PDF downloaded and verified in full (12 pages, read pp.1–10)
  • 10.1126/science.1177221 (Selman 2009) — closed-access (not_oa); partially verified via Crossref abstract and PubMed metadata only
  • 10.1007/s00125-012-2644-8 (Copps 2012) — download failed (no candidate URLs); unverifiable

Key corrections:

  • IRS-1 phosphoserine sites: removed Ser1101 (absent from Um 2006); corrected human numbering to Ser312, Ser636/639 (per Tremblay 2005 cited in Um 2006); added Ser302 (Harrington 2004 cited in Um 2006)
  • Holz 2005 footnote cell lines: “HeLa and 293T cells” → “HEK293E cells (also U2OS and HeLa)” (293T not used)
  • rpS6 phosphorylation sites (Ser235/236/240/244) flagged as not from Holz 2005; tagged unsourced
  • Selman 2009 lifespan: replaced unsupported “~9% median” with confirmed n values (n=29 KO, n=23 WT females; p<0.001 log-rank); flagged GenAge discrepancy (19% vs 9%); tagged no-fulltext-access
  • Selman 2009 footnote: updated from n=not-specified to confirmed female cohort sizes with p-value and strain
  • Um 2006 footnote: corrected model description

Final verified state: verified: true (partial scope)

[2026-05-04] verify | pathways/nf-kb.md

Page verified: pathways/nf-kb.md

Sources checked:

  • doi:10.1016/j.cell.2008.01.020 (Hayden & Ghosh 2008, Cell) — downloaded and verified in full (16 pages read)
  • doi:10.1101/gad.183434.111 (Hayden & Ghosh 2012, Genes Dev) — downloaded and verified in full (pages 1–22 read)
  • doi:10.1371/journal.pbio.0060301 (Coppé 2008, PLoS Biol) — local PDF, verified in full (15 pages read)
  • doi:10.1056/NEJMoa1707914 (Ridker 2017, NEJM — CANTOS primary) — downloaded and verified in full (12 pages read)
  • doi:10.1016/S0140-6736(17)32247-X (Ridker 2017, Lancet — CANTOS cancer analysis) — verified via PubMed abstract (full PDF not downloaded)
  • doi:10.1016/j.molmed.2010.03.003 (Freund 2010) — download failed; not directly verified (PMC accessible; tagged no-fulltext-access in existing footnote)
  • doi:10.1016/j.arr.2007.09.002 (Salminen 2008) — not_oa; not verified (#gap/no-fulltext-access)

Corrections made:

  • κB consensus sequence: GGGRNNYYCC → GGGRNWYYCC (Hayden 2012 p.204; W=A or T, not second N; 10-base consensus confirmed)
  • Coppé 2008 SASP timing: “~7 days” → “4–7 days” (paper explicitly states “requiring 4–7 d after irradiation before expressing a robust SASP”; Fig 1D)
  • Coppé 2008 attribution error: removed claim that Coppé 2008 showed “broadly suppressed by NF-κB pathway inhibition” — Coppé 2008 characterizes the SASP secretory profile but includes no NF-κB inhibition experiments. NF-κB inhibition data belongs to Freund 2010 (now correctly attributed)
  • CANTOS cancer mortality: HR 0.49 (95% CI 0.31–0.75; p=0.0009) was wrongly attributed to 150 mg dose and to the NEJM primary paper. Corrected: HR 0.49 is from the separate Lancet cancer analysis (doi:10.1016/S0140-6736(17)32247-X) and applies to the 300 mg group only. Lung cancer reduction at 150 mg is HR 0.61 (95% CI 0.39–0.97)
  • CANTOS IL-6: “~15%” → “25–43% across dose groups” (Lancet cancer paper); primary NEJM paper does not report a specific IL-6 % in the body text
  • CANTOS 300 mg significance: added note that 300 mg did not meet its prespecified multiplicity-adjusted threshold (required p<0.01058; achieved p=0.031)
  • Added new footnote [^ridker2017lancet] for the Lancet cancer paper (doi:10.1016/S0140-6736(17)32247-X)
  • Archive status updated: Hayden 2008, Hayden 2012, CANTOS NEJM all now locally downloaded
  • Removed ⚠️ auto-extraction banner; replaced with scope note

Unverifiable claims:

  • Freund 2010 (Trends Mol Med) — IL-1α autocrine loop, specific SASP timing claims: download failed, PMC accessible but not read; tagged no-fulltext-access
  • Salminen 2008 (Ageing Res Rev) — not_oa; claims about broad NF-κB activation in non-senescent aged tissues not verifiable
  • Drosophila Relish lifespan “~6–15%” claim — unsourced, no primary citation provided
  • NEMO het mouse longevity claim — unsourced
  • C. elegans nfkb-1/rel-1 claim — unsourced
  • KEGG hsa04064, Reactome R-HSA-9020702, WikiPathways WP3539 — not confirmed against live databases

Downstream propagation needed:

  • processes/sasp.md — cites Coppé 2008 and likely repeats the erroneous “NF-κB pathway inhibition” framing; the SASP timing (4–7 days vs ~7 days) should also be checked there
  • hallmarks/chronic-inflammation.md — may cite CANTOS with the wrong dose/HR attribution for cancer mortality
  • Any study page for Ridker 2017 NEJM — verify the cancer mortality claim is qualified as 300 mg / from separate Lancet paper

Final verified state: verified: true (partial scope — Freund 2010, Salminen 2008, DB IDs, and model-organism lifespan claims not verified from primary sources)

[2026-05-04] verify | interventions/pharmacological/senolytics.md

Page verified: interventions/pharmacological/senolytics.md Sources checked:

  • 10.1111/acel.12344 (Zhu 2015) — local PDF, fully verified (15 pages)
  • 10.1111/acel.12445 (Zhu 2016 navitoclax) — local PDF, fully verified (8 pages)
  • 10.1016/j.ebiom.2018.12.052 (Justice 2019) — local PDF, fully verified (10 pages)
  • 10.1016/j.ebiom.2019.08.069 (Hickson 2019) — local PDF, fully verified (11 pages)
  • 10.1038/nature16932 (Baker 2016) — local PDF, fully verified (relevant sections, 30 pages)
  • 10.1038/nm.4010 (Chang 2016) — DOI lookup failed (green OA, 0 candidate URLs after PMC filter); unverifiable
  • 10.1038/nature10600 (Baker 2011) — DOI lookup failed; unverifiable (carried from p21 verification round)
  • NCT04129944, NCT02874989, NCT02848131, NCT03675724, NCT03325322 — verified via ClinicalTrials.gov API

Key corrections:

  • HUVEC SCAP row: removed PI3Kδ — per Zhu 2015 Fig 1D-G, PI3KCD siRNA is senolytic in preadipocytes (Fig 1D) but is not among the selective HUVEC SCAP nodes (Fig 1E,G show EFNB1 and BCL-xL as HUVEC-selective). Corrected in the cell-type SCAP table and in the compound list bullet for quercetin.
  • Hickson 2019 footnote Q dose added: Q 1000 mg twice daily (not just “3-day D+Q course”). SASP factors expanded: IL-1α, IL-2, IL-6, IL-9, MMP-2, -9, -12 all decreased (wiki listed only IL-1α, IL-6, MMPs-9/-12).
  • NCT03325322 (DKD Fisetin) corrected from “ongoing” to “suspended” per ClinicalTrials.gov API.
  • Chang 2016 inline claim narrowed: “lung/liver/bone marrow senescent cell clearance” cannot be confirmed without the PDF; claim corrected to “improved HSC function” (confirmed from paper title) with no-fulltext-access tag on the broader tissue claims.
  • Chang 2016 footnote updated to note archive failure and that the title confirms HSC rejuvenation but not tissue-specific clearance claims.
  • Amor 2020 CAR-T paper confirmed real: Nature 10.1038/s41586-020-2403-9 “Senolytic CAR T cells reverse senescence-associated pathologies.”

Unverifiable claims:

  • Chang 2016 lung/liver/bone marrow senescent cell clearance data — PDF not accessible (#gap/no-fulltext-access)
  • Baker 2011 INK-ATTAC progeroid data — PDF not accessible (carried from p21 verification)
  • Amor 2020 details — DOI confirmed but PDF not read

Downstream propagation needed:

  • molecules/compounds/quercetin.md — may have the same PI3Kδ HUVEC attribution error; check and correct SCAP node table
  • Any page inheriting the “lung/liver/bone marrow” Chang 2016 claim should be qualified

[2026-05-04] verify | molecules/compounds/nr.md

Page verified: molecules/compounds/nr.md Sources checked:

  • 10.1038/ncomms12948 (Trammell 2016) — local PDF, fully verified (10 pages)
  • 10.1038/s41467-018-03421-7 (Martens 2018) — local PDF, fully verified (9 pages)
  • 10.1016/j.cmet.2022.02.001 (Brakedal 2022) — local PDF, fully verified (10 pages)
  • 10.1038/ncomms13103 (Ratajczak 2016) — local PDF, fully verified (6 pages)
  • 10.3389/fcvm.2022.881703 (Freeberg 2022) — local PDF, fully verified (5 pages; study protocol only)
  • 10.1126/science.aaf2693 (Zhang 2016) — download failed (green OA via EPFL repository inaccessible); unverifiable

Key corrections:

  • Trammell 2016 n: “~12” → “12 (clinical trial arm) + n=1 pilot”; clarified that the 2.7-fold NAD+ rise was from the n=1 preliminary experiment, not the clinical trial; added ChromaDex COI disclosure
  • Martens 2018 n: “n=24” → “n=30 randomized; 24 completed”; dose clarified as 500 mg × 2/day; NAD+ elevation precision added (median 12.2 vs 7.7 pmol/mg protein, p=0.048 one-sided); BP finding corrected — overall SBP change not statistically significant after multiple comparison correction; subgroup finding (−9 mmHg in elevated-BP group, n=13) correctly labeled as exploratory/no formal inference
  • Brakedal 2022 duration: “12 weeks” → “30 days” (critical error); patient population clarified as newly diagnosed, treatment-naive PD patients; MRS response rate added (10/13 with data showed increase, not universal); UPDRS result clarified — no significant change in overall NR group; improvement trend only in MRS-responder subgroup (n=10)
  • Freeberg 2022 recharacterized throughout: was described as a completed results study (“Follow-up study… Specifically evaluated… Provides mechanistic follow-through”); corrected to study protocol paper (NCT03821623; n=94 planned; no results reported at publication); Limitations section corrected accordingly
  • Ratajczak 2016 model description: “in-vitro / mammalian cells” → “in-vitro + in-vivo (mammalian cell lines + mouse IP injection)”
  • PK table: “~40–60% above baseline” replaced with exact Martens 2018 values; “Duration to steady-state ~14 days” removed (not stated in paper); Brakedal 2022 brain NAD+ entry corrected to 30 days and added responder fraction
  • Safety section: “up to 12 weeks” corrected to “up to 6 weeks (Martens) / 30 days (Brakedal)”; AE characterization corrected (Martens dropouts were in placebo phase, not NR phase)
  • Zhang 2016 lifespan claim: “+4.6% in females” tagged as unverified (PDF inaccessible); citation count updated to 1173 from DOI lookup
  • Banner removed; verified: true (partial scope — Zhang 2016 and canonical DB fields not re-checked)

Unverifiable claims:

  • Zhang 2016 (Science) lifespan extension percentage and sex breakdown — PDF download failed; no-fulltext-access; recommend retry via institutional access or author manuscript when available

Downstream propagation needed:

  • studies/brakedal-2022-nadpark-parkinsons.md — if it exists, duration should be corrected to 30 days
  • studies/martens-2018-nr-older-adults.md — if it exists, n should be 30 randomized / 24 completed; BP result should be corrected
  • studies/freeberg-2022-nr-blood-pressure.md — if it exists, should be recharacterized as a protocol paper
  • hallmarks/deregulated-nutrient-sensing.md and hallmarks/mitochondrial-dysfunction.md — may cite Brakedal 2022 with wrong duration

[2026-05-04] verify | molecules/compounds/metformin.md

Page verified: molecules/compounds/metformin.md Sources checked: Martin-Montalvo 2013 (10.1038/ncomms3192) — local PDF, fully verified. Coll 2019 (10.1038/s41586-019-1911-y) — downloaded, verified. Barzilai 2016 (10.1016/j.cmet.2016.05.011) — downloaded, verified. Anisimov 2008 (10.4161/cc.7.17.6625) — abstract verified via EuropePMC (PDF download failed: no candidate URLs). Owen 2000 (10.1042/bj3480607) — downloaded file is author index page only, not full paper text; claims unverifiable from local copy. Bannister 2014 (10.1111/dom.12354) — not_oa, unverifiable. ChEMBL1431 and PubChem CID 4091 confirmed via live API; DrugBank DB00331 confirmed via PubChem cross-reference. Key corrections:

  • Martin-Montalvo 2013 primary lifespan arm: organism corrected C57BL/6 (not B6C3F1); n corrected to n=64 metformin / n=83 SD controls (not “~60/group”); treatment start age corrected to 54 weeks (~13.5 months), not “3 months”; p-value and test name (Gehan-Breslow, p=0.02) added; B6C3F1 replication arm explicitly distinguished (p=0.064, NS; n=36/297).
  • Anisimov 2008 lifespan extension corrected from “+8%” to “+37.8%” mean lifespan; organism description corrected from “spontaneously hypertensive rats” to “female outbred SHR mice.”
  • B12 deficiency claim (~19% reduction) removed from martinmontalvo2013 citation (not in that paper); re-attributed to barzilai2016 with correct figure (7% vs 5% incidence over 13 years in DPP/DPPOS).
  • CR+Met combination lifespan claim (“did not produce additive extension”) removed — Martin-Montalvo 2013 did not include a combined Met+CR lifespan arm; CR group was for transcriptome comparison only.
  • Coll 2019 GDF15 section: added that glucose-lowering is GDF15-independent (Gdf15-null mice retain insulin sensitivity response); clarified GFRAL receptor location (hindbrain, not just “area postrema”); noted actual publication year Nature 578:444–448, 2020.
  • Footnote for martinmontalvo2013 updated with correct n and p-value details.
  • Banner removed; verified: true (partial scope per verified-scope field). Unverifiable claims: Owen 2000 complex I mechanism detail (downloaded file not readable as paper); Bannister 2014 HR figures (not_oa). ITP 2024 Gehan reanalysis claim is not independently sourced. unsourced on that specific sub-claim. Downstream propagation needed: studies/martin-montalvo-2013-metformin-mice.md (if it exists) should be updated with correct n, treatment age, and p-value. studies/anisimov-2008-metformin-shr.md should be corrected if it exists. Any MOC pages (interventions/pharmacological/geroprotectors, hallmarks/deregulated-nutrient-sensing) citing metformin lifespan data may have inherited the +8% Anisimov error.

[2026-05-04] verify | molecules/proteins/bax.md

Page verified: molecules/proteins/bax.md Sources checked:

  • 10.1016/0092-8674(93)90509-o (Oltvai 1993) — not_oa; author list and title verified via Crossref.
  • 10.1126/science.270.5233.96 (Knudson 1995) — not_oa; author list, title, and page range (96–99) verified via Crossref. Neuronal excess claim not independently verifiable without full PDF; confirmed via Chipuk 2010 review summary.
  • 10.1126/science.1092734 (Chipuk 2004) — not_oa; mechanism claim (cytosolic p53 directly activates BAX) verified via Crossref abstract.
  • 10.1016/j.molcel.2010.01.025 (Chipuk 2010) — downloaded and read in full (PDF). Review title confirmed “The BCL-2 Family Reunion”; PUMA classification nuance verified; p53 as non-BCL-2-family direct activator confirmed; both direct-activation and de-repression models confirmed as current consensus. pp. 299–310.
  • 10.18632/aging.101202 (Zhu 2017) — downloaded and read in full (PDF). Critical error found: study is in vitro only.
  • 10.1111/acel.14229 (Shen 2024) — downloaded and read in full (PDF). n=3 per group confirmed; BTSA1 dose 10 mg/kg IP confirmed; minority MOMP mechanism confirmed.
  • UniProt Q07812 — domain residues (BH3: 59–73, BH1: 98–118, BH2: 150–165, TM: 172–192) and PTMs (Met1-acetylation, Lys128/Lys190-ubiquitination) confirmed via REST API.
  • GenAge Build 21 — BAX confirmed absent from human database.

Key corrections:

  • Zhu 2017 in-vivo claim → corrected to in-vitro only. Wiki had ”[^zhu2017]” labeled “in-vitro + in-vivo”; Zhu 2017 is entirely in vitro (HUVECs, IMR90, preadipocytes). No in vivo arm. Body text and footnote corrected.
  • Zhu 2017 drug attribution corrected: wiki implied navitoclax/ABT-263 was the primary drug tested; Zhu 2017 primarily tests A1331852 and A1155463 (selective BCL-xL inhibitors). Navitoclax is a comparator from prior work.
  • Zhu 2017 cell-type specificity added: A1331852 and A1155463 senolytic in HUVECs and IMR90 but NOT preadipocytes; fisetin senolytic in HUVECs only.
  • Shen 2024 n=N/A → n=3 per group (confirmed from methods/figures). BTSA1 dose (0.3 µM in vitro; 10 mg/kg IP in vivo) added.
  • Chipuk 2010 citation count “1,398” removed (stale at time of extraction; Crossref shows 1,251 as of 2026-05-04; live metric removed rather than frozen).
  • GenAge gap tag updated: confirmed BAX absent from GenAge human database Build 21.
  • Banner removed; verified: true with scope.

Unverifiable claims:

  • Knudson 1995 neuronal specificity (facial motor nucleus) — plausible given Chipuk 2010 review context, but full PDF is closed-access; not independently verified line-by-line.
  • Oltvai 1993 “21 kDa” molecular weight — stated in wiki, plausible, closed-access.
  • Citation counts in oltvai1993 footnote (5,984 citations) — live metric not re-verified.

Downstream propagation needed:

  • Any study pages citing Zhu 2017 that characterize it as in vivo should be corrected.
  • interventions/pharmacological/senolytics.md (if it exists) may repeat the in-vivo claim for A1331852/A1155463.
  • molecules/compounds/ pages for navitoclax, A1331852, A1155463, fisetin, BTSA1 if they cite Zhu 2017 or Shen 2024 should be checked.

[2026-05-04] verify | molecules/proteins/bcl-2.md

Page verified: molecules/proteins/bcl-2.md Sources checked:

  • 10.1111/acel.12344 (Zhu 2015) — local PDF, fully read and verified
  • 10.1111/acel.12445 (Zhu 2016) — local PDF, fully read and verified
  • 10.1056/nejmoa1513257 (Roberts 2015, NEJM) — downloaded and read in full
  • 10.3389/fcell.2020.00354 (Sharma 2020) — downloaded and read in full
  • UniProt P10415 — caspase cleavage site and domain residues cross-checked via REST API
  • 10.1038/nm.4010 (Chang 2016) — download failed (green OA, no PMC entry); unverifiable
  • 10.1073/pnas.82.21.7439 (Cleary 1985) — not_oa; unverifiable
  • 10.1038/335440a0 (Vaux 1988) — not_oa; unverifiable

Key corrections:

  • IMR90 SCAP claim overstated: “BCL-2 and BCL-w are the dominant SCAP nodes” → corrected to “triple BCL-2 + BCL-xl + BCL-w siRNA combination required for IMR90 senolysis” per Zhu 2016 Fig. 4A–B. BCL-xl alone and BCL-2 + BCL-xl together were insufficient in IMR90; BCL-w co-knockdown was required. BCL-xL alone sufficient in HUVECs.
  • Caspase cleavage site corrected: “between BH3 and BH4 domains” → “between BH4 and BH3 domains (at position 34)” per UniProt P10415. BH4 spans 10–30, BH3 spans 93–107; cleavage at position 34 is between BH4 and BH3, not between BH3 and BH4.
  • Roberts 2015 enrollment criterion corrected: “relapsed/refractory CLL with 17p deletion” → “relapsed/refractory CLL or SLL; 17p deletion present in 30% of CLL patients but not an enrollment requirement.” 79% overall response rate figure confirmed (95% CI 71–86) across all 116 patients.
  • Roberts 2015 trial phase corrected: “phase 1/2” → “phase 1 dose-escalation” (as stated in the paper’s methods).
  • Sharma 2020 bone loss claim quantified: added −60.1% (females) / −45.6% (males) trabecular bone volume fraction decrease and −88% / −83% mineralized matrix production impairment; dose specified as 50 mg/kg/day oral gavage for 2 weeks.
  • Venetoclax senolytic context updated: BCL-2-only inhibition predicted insufficient for IMR90 fibroblast senolysis based on Zhu 2016 triple-knockdown requirement.
  • Chang 2016 footnote updated: marked no-fulltext-access; claims noted as unverified.
  • Zhu 2016 footnote corrected: figure reference updated from “Fig. 5B” to “Fig. 4A–B”; IMR90 claim corrected.
  • Pathway membership section corrected to match verified IMR90 SCAP language.
  • Banner removed; verified: true (partial scope per verified-scope field).

Unverifiable claims:

  • Chang 2016 (navitoclax HSC/MuSC rejuvenation in aged mice) — download failed; no PMC entry. Marked no-fulltext-access.
  • Cleary 1985 and Vaux 1988 historical claims — both not_oa; retained as-is with existing notation.
  • Canonical-database identity fields (UniProt accession, NCBI Gene 596, HGNC 990, Ensembl ENSG00000171791, GenAge ID 69) not independently re-verified against databases.

Downstream propagation needed:

  • studies/zhu-2016-navitoclax-senolytic.md (if it exists) — may have the old “BCL-2 + BCL-w dominant in IMR90” claim; needs correction to triple-knockdown finding.
  • interventions/pharmacological/senolytics.md — may repeat the BCL-2/BCL-w phrasing; update to triple BCL-2/BCL-xl/BCL-w for IMR90.
  • Any page citing Roberts 2015 with “17p deletion” as enrollment criterion should be corrected to the broader CLL/SLL population.
  • molecules/compounds/navitoclax.md (if it exists) — check bone safety claim for dose specificity; check IMR90 SCAP language.

[2026-05-04] verify | pathways/sirtuin.md

Page verified: pathways/sirtuin.md Sources checked: Tissenbaum 2001 (local PDF, verified), Imai & Guarente 2016 (downloaded, verified), Pacholec 2010 (downloaded, verified), Hubbard 2013 (downloaded, verified). Closed-access: Imai 2000, Verdin 2015, Garten 2015, Kanfi 2012, Pencina 2023 — all not_oa; unverifiable. Corrections: No numeric corrections required — all quantitative claims from the 4 verified sources matched the PDFs exactly. Footnotes for all 4 verified sources enriched substantially: Tissenbaum 2001 (n per strain added: 80–451; specific mean lifespan values added; daf-16 epistasis confirmed); Pacholec 2010 (Pfizer attribution added; native substrate results for Ac-p53 and Ac-AceCS1 specified; SRT1720 in vivo null result added; target promiscuity data added); Hubbard 2013 (substrate identities PGC-1α K778 and FOXO3a K290 specified; murine E222K equivalent noted; 117-compound panel confirmed; cell model specified). no-fulltext-access tags applied to all load-bearing claims from closed-access sources (Verdin 2015 three-mechanism framing; Kanfi 2012 lifespan/IGF-1 claims; Pencina 2023 RCT parameters). Banner updated; verified: true (partial scope). Downstream propagation needed: None currently. SIRT1/SIRT3/SIRT6 protein pages (not yet seeded) should inherit the Pacholec/Hubbard framing when created.

[2026-05-04] verify | rapamycin compound page

Page verified: molecules/compounds/rapamycin.md Sources verified: Harrison 2009 (local), Miller 2014 (downloaded), PEARL 2025 (local), Arriola Apelo 2016 (downloaded). Mannick 2014, Johnson 2013 review, Lamming 2012 — unverifiable (not_oa or download failed). Key corrections: Harrison 2009 dose (14 mg/kg food not body weight), n (1,960 total; 674 controls), mouse cross corrected; Miller 2014 n corrected (156/136 per rapamycin group, not 40-60) and three doses documented; PEARL n corrected (114 completers, not ~50), primary endpoint VAT was non-significant (p=0.942), bioavailability caveat for compounded rapamycin added; Arriola Apelo model specificity added (male C57BL/6J only). Banner removed; verified: true (partial scope).

[2026-05-04] verify | pathways/apoptosis-pathway.md

PDFs read: Tyner 2002 (local), Zhu 2015 (downloaded gold OA), Zhu 2016 (downloaded gold OA), Danial & Korsmeyer 2004 (downloaded bronze OA). Lossi 2022 closed-access; claims attributed to it backed by Danial 2004.

Corrections (5 substantive): (1) p44/Δ40p53 mis-attribution removed — Tyner 2002 m-allele is chimeric C-terminal fragment not p44; (2) stem-cell depletion corrected from “demonstrated” to “hypothesized”; (3) Tyner n corrected to n=35+56 headline, total ~480, with verified lifespan numerics; (4) BCL-xL “dominant SCAP” claim corrected to cell-type-specific pattern per Zhu 2015 Fig. 5; (5) navitoclax aged-mouse lung/liver/bone-marrow/HSC claim removed — not from Zhu 2016, which is in vitro + progeroid MEFs only.

Downstream propagation needed: quercetin.md, dasatinib.md (BCL-xL claim + Zhu 2015 SCAP framing); any page citing Zhu 2016 for aged-mouse navitoclax data.

[2026-05-04] verify | p21 (CDKN1A)

Page verified: molecules/proteins/p21.md

Sources verified:

  • Brugarolas 1995 (10.1038/377552a0) — local PDF, verified
  • Tyner 2002 (10.1038/415045a) — local PDF, verified (also cross-checked against verified study page)
  • Baker 2016 (10.1038/nature16932) — downloaded and verified
  • Blanc 2016 (10.1016/j.celrep.2016.01.022) — downloaded and verified
  • Dimri 1995 (10.1073/pnas.92.20.9363) — downloaded and verified
  • UniProt P38936 — REST API verified
  • GenAge 284 — live database verified
  • Baker 2011 (10.1038/nature10600) — OA green, download failed twice; unverifiable no-fulltext-access
  • el-Deiry 1993, Harper 1993, Waga 1994 — closed access; not verified

Corrections made:

  1. Wrong attribution — G2/M checkpoint: Brugarolas 1995 was cited as evidence that p21-null MEFs demonstrate p21’s role in G2/M arrest. The paper actually shows p21-null MEFs have an intermediate G1 arrest defect (S-phase fractions 57.5–54.5% of untreated vs 39% for WT); it does not address G2/M. Citation note added; G2/M in-vivo claim tagged unsourced.
  2. Checkpoint table G2/M row: “yes (partial; CDC25C is co-required)” → “partial/unsourced in vivo unsourced”
  3. Wrong attribution — Dimri 1995 cited for p21 induction in aged human skin: Dimri 1995 establishes SA-β-Gal as a senescence biomarker; it does not measure or discuss p21 protein. Attribution corrected; p21-specific claim tagged unsourced; footnote updated.
  4. Missing PTM sites (UniProt): UniProt P38936 lists Ser-114 (GSK3β), Ser-130, and Ser-160 (PKC) as phosphorylation sites not in the wiki PTM table. Added as a verification note; table expansion flagged for next pass.
  5. Baker 2016 footnote enriched: Added confirmed lifespan figures (27% mixed, 24% C57BL/6; ranges 17–35% by sex/background), strain background details, and clearance selectivity notes.
  6. Brugarolas 1995 footnote corrected: Updated from “in-vivo (mouse, p21-null MEFs)” to “in-vitro (MEFs)” with specific S-phase fractions and explicit note that G2/M checkpoint is not addressed.
  7. Dimri 1995 footnote corrected: Added explicit warning not to cite for p21 induction; described actual SA-β-Gal finding and in vivo skin data (20 donors aged 20–90 yr).
  8. Blanc 2016 footnote corrected: Added specific quantitative findings (~30% SA-β-Gal in PRMT7-KO; ~22% in conditional KO; DNMT3b rescue ~65%); clarified model system.
  9. Baker 2011 footnote updated: Marked as OA-but-download-failed with no-fulltext-access.
  10. HGNC field: Contains symbol (CDKN1A) not numeric ID (1784); flagged in frontmatter comment.
  11. Banner removed; verified flags flipped to true.

Unverifiable claims (closed-access):

  • el-Deiry 1993 (WAF1 identification), Harper 1993 (CIP1 identification), Waga 1994 (PCNA-binding mechanism) — all closed access; kept with existing gap tags.

Downstream propagation needed:

  • pathways/p53-pathway.md — if it cites Brugarolas 1995 for G2/M checkpoint evidence, that citation needs the same correction applied here.
  • Any page citing dimri1995 for p21 protein levels in aged human skin needs the attribution corrected.

[2026-05-04] seed+verify | Round 3b — Bcl-2 family proteins (4 of 4 drafted + verified)

Bcl-2 family proteins drafted (parallel seeders, ~5 min) then verified (parallel verifiers, ~10 min):

  • molecules/proteins/bcl-xl.md (verified-partial; 10 corrections from verifier)
  • molecules/proteins/bcl-2.md (verified-partial; 5 corrections incl. major IMR90 SCAP triple-knockdown finding)
  • molecules/proteins/bax.md (verified-partial; Zhu 2017 in vitro-only correction)
  • molecules/proteins/puma.md (verified-partial; 7 corrections incl. Baker 2016 wrong-attribution for SCAP upregulation)

Cross-cutting Round 3b findings (the SCAP framework needed substantial refinement):

  • IMR90 SCAP framing was wrong across multiple pages. The simple “BCL-2 + BCL-w dominant in IMR90” framing (which I had used in senolytics.md based on Zhu 2015 Fig 5B summary) is incomplete. Two papers refine it:
    • Zhu 2016 Fig 4A-B: triple BCL-2 + BCL-xL + BCL-w knockdown required for IMR90 senolysis; BCL-xL alone or BCL-2+BCL-xL together were insufficient
    • Yosef 2016: BCL-W + BCL-xL dual knockdown produced ~53% viability reduction
    • Combined: IMR90 senolysis requires multi-target Bcl-2 family inhibition; no single member is dominant
  • Preadipocyte SCAP framing: ephrins/PI3KCD/p21/PAI-2 (not BCL-2/BCL-xL secondary). Dasatinib’s mechanism is correctly framed; senolytics.md page already had this right.
  • Zhu 2017 was in vitro only (wiki had claimed in vivo); the in vivo navitoclax/aged-mouse work is Chang 2016 Nat Med (correctly attributed in senolytics.md from earlier propagation)
  • Yosef 2016 in vivo was targeted senescence in YOUNG mice, not aged-mouse healthspan
  • Baker 2016 doesn’t address SCAP upregulation — it’s the INK-ATTAC lifespan paper. Wrong attribution removed from puma page.
  • Schema addition: sens-categories field added to protein frontmatter (analogous to pathway schema; useful for proteins like BCL-xL that map to ApoptoSENS)
  • Multiple wrong author/year/cell-line errors caught by puma verifier (Nakano 2001 cell lines; Yu 2001 missing Hwang as third author; FOXO3a-Yu attribution wrong)

Propagation pass applied this round:

  • interventions/pharmacological/senolytics.md § The SCAP framework — IMR90 row in cell-type table updated to “Multi-target Bcl-2 family — no single member sufficient”; new explanatory note added with Zhu 2016 / Yosef 2016 citations and the practical implication (navitoclax suited; single-target agents like venetoclax not)

Outstanding propagations (deferred to future lint):

  • pathways/p53-pathway.md — possibly carries FOXO3a + BCL-w/MCL-1/A1 misattributions per puma verifier
  • processes/apoptosis-pathway.md — Yosef 2016 framing may need check
  • molecules/compounds/quercetin.md and dasatinib.md — already verified, but BCL-xL SCAP claim restriction may need a sweep

Final state:

  • 31 entity pages now verified: true (4 added this round)
  • 1 entity page verified: false — senolytics.md (drafted by me in main thread; verifier pass dispatched)

[2026-05-04] verify | molecules/proteins/ulk1.md

Page verified: molecules/proteins/ulk1.md

Sources checked:

  • 10.1038/ncb2152 (Kim 2011) — local PDF, fully verified (10 pages)
  • 10.1091/mbc.e08-12-1248 (Hosokawa 2009) — downloaded green OA, fully verified (11 pages)
  • 10.1074/jbc.m900573200 (Ganley 2009) — downloaded hybrid OA, fully verified (9 pages)
  • 10.1038/ncomms3300 (Pyo 2013) — downloaded green OA, fully verified (8 pages)
  • 10.14336/AD.2025.0419 (Zhang 2025) — downloaded gold OA, fully verified (8 pages)
  • 10.1038/s41580-018-0033-y (Hansen 2018) — download failed on two attempts (green OA URL unreachable); unverifiable; claims tagged no-fulltext-access
  • UniProt O75385 — domain ranges and PTM sites cross-checked via REST API 2026-05-04

Key corrections:

  • Ser556 PTM table: “AMPK (also mTOR)” → “AMPK” only. The “also mTOR” claim for human ULK1 Ser556 is not supported by Kim 2011 (which covers mouse Ser317/Ser777 and Ser757 only) or by UniProt O75385 annotations. Tagged unsourced.
  • ATG101 as core complex subunit: citation [^hosokawa2009][^ganley2009] removed — neither paper describes ATG101. Both characterize only the ULK1–ATG13–FIP200 trimer. ATG101 claim tagged unsourced; note added that ATG101 was identified as a fourth subunit in subsequent work (Mercer et al. and others).
  • Hosokawa 2009 footnote corrected: added explicit note that neonatal lethality was NOT reported in Hosokawa 2009 (MEF autophagy defects only); lethality is from Kundu et al. 2008 (Blood 112:1493–1502).
  • Ulk1−/− lethality in body text: citation [^hosokawa2009] removed; claim tagged unsourced with pointer to Kundu 2008 (doi:10.1182/blood-2008-01-132639); correct citation needed.
  • Pyo 2013 lifespan figure: “~17%” → “17.2%” (precise figure from abstract; χ²=17.32, p<0.001 log-rank). Footnote enriched with n=65 WT + 70 Tg, mouse line no. 25, pCAGGS promoter, C57BL/6 background, and additional phenotypes (leanness, insulin sensitivity).
  • Ganley 2009 footnote cell lines: “HEK293 and NIH3T3” → “MEF and 293T cells” (per paper methods).
  • Hansen 2018 footnote: marked no-fulltext-access; banner body-text claim tagged accordingly.
  • Zhang 2025 footnote: clarified that paper is a secondary review with no primary exercise muscle data; sites cited (Ser317, Ser555) are from review of rodent literature, not original measurements.
  • Hosokawa 2009 and Ganley 2009 footnotes: updated download status to “locally downloaded.”
  • Pyo 2013 footnote: updated download status to “locally downloaded.”
  • Limitations table: added three new rows for (1) Ser777 quote from Kim 2011 Discussion, (2) ATG101 citation gap, (3) Ulk1−/− lethality citation gap.
  • Kim 2011 species-numbering caveat note: confirmed accurate against Discussion of Kim 2011 (quote: “Ser 777 in the mouse Ulk1 is not conserved in human Ulk1”).
  • ⚠️ banner removed; verified flag flipped to true (partial scope — Hansen 2018 unverifiable).

Unverifiable claims:

  • Hansen 2018 claim that Atg5/Atg7 RNAi abolishes rapamycin lifespan extension in worms/flies — PDF not accessible; tagged no-fulltext-access.

Downstream propagation needed:

  • pathways/autophagy.md (if it exists) — may cite ATG101 as a ULK1-complex subunit without the correct primary citation; check and tag.
  • pathways/mtor.md or pathways/ampk.md — may inherit the Ser556 “also mTOR” claim; check PTM tables.
  • Any study pages for Hosokawa 2009 or Ganley 2009 — if they exist, should not list ATG101 in their key-findings without the correct citation.
  • studies/pyo-2013-atg5-lifespan.md — if it exists, update lifespan to 17.2% and n values.

[2026-05-04] seed | Round 3a — senolytics category page (first category-page prototype)

Drafted interventions/pharmacological/senolytics.md (~250 lines, verified: false) in main thread. This is the first category page — distinct from individual compound pages by aggregating across the drug class rather than diving deep on one compound.

Schema decisions made (sets convention for future drug-class pages):

  • Same type: intervention schema as caloric-restriction.md with mode: pharmacological.
  • mechanisms: field lists the class-level SCAP-disruption mechanisms (rather than specific molecular targets, which live on compound pages).
  • target-hallmarks: and target-pathways: populated normally.
  • Body structure: definitional clarifications (senolytic vs senomorphic vs senotherapeutic vs hit-and-run) → SCAP framework → drug families table (BH3-mimetics / TKI / flavonoids / other) → clinical evidence aggregate table → cell-type specificity discussion → SENS/hallmark mapping → related approaches (genetic clearance, senomorphics, CAR-T) → limitations.
  • Each compound mention links to the molecules/compounds/ page when one exists; planned/implicit-stub compounds (navitoclax, a1331852) noted with (planned).

Content highlights:

  • SCAP framework directly inherits the cell-type specificity corrections from Zhu 2015 verifier rounds — HUVECs (BCL-xL/PI3Kδ/EFNB1), IMR90 (BCL-2/BCL-w), preadipocytes (ephrins/EPH/PI3KCD/serpins/p21).
  • Chang 2016 Nat Med correctly attributed for navitoclax aged-mouse senolytic data — explicit note that Zhu 2016 (which is in vitro only) is a common misattribution. This is the propagation fix from the apoptosis-pathway verifier round.
  • UNITY-OA-101 Phase 2 failure documented as the first major senolytic clinical setback.
  • Baker 2016 lifespan extension figures (27% mixed bg / 24% C57BL/6) from the p21 verification round inherited correctly.
  • Justice 2019 + Hickson 2019 numbers match the verified dasatinib/quercetin/apoptosis-pathway pages.

Implicit stubs newly created: navitoclax, a1331852, ubx0101 (BH3-mimetic family); foxo4-dri-peptide; some others.

ROADMAP.md updated: interventions/pharmacological/senolytics flipped from [ ] (9 refs) to [x] (drafted). This was the top single missing entity per the post-Round-2 stub queue.

Next: Round 3b candidates (Bcl-2 family proteins, mTOR-cluster proteins, NAD+ family compounds) can be dispatched as parallel subagent batches.

Graph View

Backlinks