R12

log/R12.md — Round 12 entries

Sub-file of log — see parent for index.

[2026-05-05] Round 13 batch summary (post-R12 lint highest-leverage adds)

Seeders (8 in parallel, all completed)

After R12 the post-lint queue surfaced these as the highest-priority unblocks. User guidance: fold lamp-2a into a unified lamp2 atomic page; main agent’s call on hsp70 (chose family page covering inducible HSPA1A/HSPA1B/HSPA1L, distinct from existing constitutive hsc70).

• framework: frameworks/apoptosenes.md — 18 inbound (SENS damage category MOC; navigational)
• protein: molecules/proteins/pgc-1alpha.md — 13 inbound (master mitochondrial biogenesis coactivator)
• protein: molecules/proteins/il-1b.md — 12 inbound (CANTOS target; canakinumab-direct)
• pathway: pathways/scfa-signaling.md — 11 inbound (SCFA biology — wiki’s first deep-dive into microbiome metabolite signaling)
• protein: molecules/proteins/lamp2.md (folds lamp-2a alias; 10 inbound) — covers all 4 splice isoforms
• protein: molecules/proteins/tet2.md — 9 inbound (CHIP driver / dioxygenase)
• protein family: molecules/proteins/hsp70.md — 8 inbound (inducible HSPA1A/HSPA1B/HSPA1L; companion to constitutive hsc70 verified-partial R10d)
• pathway: pathways/mitochondrial-dynamics.md — 8 inbound (synthesis of drp1+mitofusins+opa1 cluster)

Verifiers (8 in parallel, all completed)

Major cross-cutting corrections:

• Wenz 2009 RETRACTED (caught by pgc-1alpha seeder): the 2009 PNAS muscle-Pgc-1α-OE-extends-lifespan paper was retracted December 2016. Brief originally cited it as a key aging-muscle citation; replaced with Leick 2010 (Exp Gerontol) at seed time. Schema escalation: CLAUDE.md does not currently document a retracted-paper citation convention. Seeder approach (bold “RETRACTED” prefix in footnote with retraction DOI) is documented for user adoption.
• PGC-1α Akt phosphorylation site CRITICAL CORRECTION (pgc-1alpha verifier): wiki claimed S265 + S568; Li 2007 PDF actually identifies Ser570 as dominant Akt site (S570A mutation markedly reduces Akt-induced effects on hepatic gluconeogenesis). S265 and S568 do not appear as Akt sites in Li 2007. PGC-1α PTM table fully corrected; sirt1.md and ampk.md cross-references propagated.
• PGC-1α GCN5 acetylation site numbering (pgc-1alpha verifier): Rodgers 2005 uses mouse sequence throughout (K77/K144/K183/K253/K270/K277/K320/K346/K412/K441/K450/K757/K778); human UniProt Q9UBK2 numbering offset is +2. Dual-column table now explicit. K757 (mouse) / K758 (human) was missing from body table at seed time; added.
• Lin 2004 Pgc-1α-null phenotype CORRECTED (pgc-1alpha verifier): wiki claimed “neurodegeneration in older mice”; paper actually shows spongiform striatal lesions in 3-month-old mice (NOT aged). Cardiac dysfunction claim was unsourced and removed (not in Lin 2004). Quantitative endpoints added: body temperature drops to ~33.5°C within 3h at 4°C; UCP1 ~45% WT in BAT; 30-60% reduction in mitochondrial gene expression in quadriceps.
• AMPK PGC-1α phosphorylation site #gap/unsourced RESOLVED on sirt1.md (R13 propagation): Jager 2007 (10.1073/pnas.0705070104) now properly cited via cross-link to verified pgc-1alpha.md. Mouse-vs-human numbering note (Thr177/Ser538 mouse = T178/S539 human) propagated to ampk.md and sirt1.md.
• CBP/p300 PGC-1α writer attribution caveat (pgc-1alpha + cbp-p300 verifiers): Rodgers 2005 demonstrates SIRT1 as the eraser but not CBP/p300 as the writer. Lerin 2006 establishes GCN5/KAT2A as the actual writer. CBP/p300-as-writer remains inferential and is flagged accordingly on pgc-1alpha.md and cbp-p300.md (R12).
• lamp-2a alias fold-in CONFIRMED (lamp2 seeder + verifier): all 10 [[lamp-2a]] inbound wikilinks now resolve to lamp2.md via Obsidian frontmatter alias. Page covers all 4 splice isoforms (LAMP-2A/2B/2C/2D).
• Zhang 2008 LAMP-2A overexpression strain CRITICAL CORRECTION (lamp2 verifier): wiki claimed C57BL/6 + “TF-LAMP2A” model name; PDF shows FVB background + Alb-Tet-off-LAMP-2A model. Age range 22-24 months → 22-26 months. Phenotype claims “albumin secretion + glucose homeostasis” REMOVED — actual measured outcomes are zoxazolamine clearance, ALT, TUNEL/caspase-3, carbonyl proteins, polyubiquitin EM ultrastructure.
• LAMP-2A 36h half-life claim REMOVED (lamp2 verifier): wiki attributed to Bandyopadhyay 2008; paper does not state a half-life figure. Tagged unsourced.
• Cuervo & Dice 2000 — Science→JBC + rat age 24→22 month (lamp2 verifier): wiki cited the wrong journal at seed time; corrected to JBC (10.1074/jbc.M002102200). Also: paper studied late-passage human fibroblasts in addition to rat liver — finding is NOT rat-only. Extrapolation table corrected.
• Bandyopadhyay 2008 700-kDa method scope (lamp2 verifier): “BN-PAGE” → 4 independent methods (native gel, BNE, gel filtration, sucrose gradient). Strengthens evidence quality.
• Nishino 2000 vs Tanaka 2000 DOI disambiguation (lamp2 verifier): 35022604 = Nishino human Danon disease; 35022595 = Tanaka Lamp2-/- mouse — both real papers in same Nature issue 406:902-906.
• Hoffman 2001 mutation names CRITICAL CORRECTION (il-1b verifier): wiki had “L353P / R260W type mutations”; paper actually identifies A439V, V198M, E627G (FCAS) + A352V (MWS) in 3 FCAS + 1 MWS families (4 total, not ~10). All four in exon 3 of CIAS1. R260W was Agostini 2004 (single MWS patient, separate study) — already correctly attributed on nlrp3-inflammasome.md.
• Heneka 2013 NLRP3-AD mechanism qualified (il-1b verifier): wiki framing made IL-1β look like the standalone M1 driver; paper shows NLRP3/caspase-1 axis drives M1→M2 skew, IL-1β is one output. Specific endpoints added: 70% FA-extractable Aβ reduction; 16-month animals; NOS2 reduction + arginase-1 + FIZZ1 increase.
• Il1b-/- KO unsourced RESOLVED (il-1b verifier): canonical sources identified as Zheng 1995 Immunity (10.1016/1074-7613(95)90154-x) and Horai 1998 JEM (10.1084/jem.187.9.1463). Two new footnotes added.
• Donohoe 2011 colonocyte energy claim CORRECTED (scfa-signaling verifier): “60-70% of energy from butyrate” was misattributed to Donohoe 2011; paper actually measures NADH/NAD+ ratios + OXPHOS rescue (~30%→70% on butyrate). The “60-70%” figure originates from Roediger 1980 Gut — flagged for citation; unsourced.
• Kimura 2011 GPR41-/- phenotype CORRECTED (scfa-signaling verifier): wiki said “reduced body fat + altered energy expenditure”; paper’s primary finding is reduced resting heart rate (682 → 610 bpm) and reduced SNS outflow. Adipose Gpr41 expression NOT detected by qRT-PCR — contradicts prior reports. Mechanism Gβγ-PLCβ-MAPK (not generic Gi-coupled).
• Smith 2013 SCFA delivery + primary ligand CORRECTED (scfa-signaling verifier): wiki said “SCFA gavage”; paper uses drinking water at 150 mM. “Propionate and butyrate most active” → propionate is the primary Ffar2/GPR43 ligand; HDAC6/9 downregulation is Ffar2-dependent (downstream of GPR43, not parallel).
• Claesson 2012 SCFA-producer attribution (scfa-signaling verifier): paper main text names Coprococcus + Roseburia (Lachnospiraceae) as depleted in long-term care, NOT Faecalibacterium prausnitzii (which is in supplementary/review framing). Cohort detail added: community 83 / day-hospital 20 / rehab 15 / long-term care 60.
• Arpaia 2013 mechanism CORRECTED (scfa-signaling verifier): butyrate drives colonic Treg only via local (enema/luminal) delivery, not systemic; propionate drives peripheral splenic Treg accumulation; acetate inactive. Wiki previously conflated all 3 papers — now distinguished.
• TET2 disease frequency table FULLY REPLACED (tet2 verifier): wiki attributed CMML ~50-60%, MDS ~25-35%, AML ~10-15% to Delhommeau 2009; actual Delhommeau 2009 figures: overall 15% (46/309), MDS 19%, MPN 12%, secondary AML 24%, CMML 22% (n=2/9 only — small subgroup). Higher CMML/MDS frequencies are from later studies. Attribution corrected.
• Moran-Crusio 2011 “~50% myeloid malignancy by 12 months” REMOVED (tet2 verifier): not in paper. Actual primary endpoint: CMML-like disease by 20 weeks (splenomegaly >250 mg, monocytosis, dysplasia). Strain (C57BL/6 × 129/SvEv hybrid Vav-Cre) and n=15 added.
• Ko 2013 IDAX DOI CRITICAL CORRECTION (tet2 verifier): brief 10.1016/j.molcel.2013.05.004 resolves to BMP-Smad paper (Xu 2013). Correct DOI: 10.1038/nature12052 (Ko M et al., Nature 2013).
• Cimmino 2017 + Agathocleous 2017 descriptions corrected (tet2 verifier): replaced approximate framing with actual primary findings (VTA-shTet2 reversible RNAi, 2-fold 5hmC increase, 30% 5mC decrease, P5 re-plate block; HSC ascorbate 2-20× normal; Gulo-/- HSC expansion).
• Frank 2001 + Karbowski 2002 BOTH DOWNLOADED + verified end-to-end (mitochondrial-dynamics verifier): bronze OA worked. Quantitative findings added: Frank 2001 — 77% TUNEL+ in DRP1-WT vs 19% in K38A at 10h (COS-7); Karbowski 2002 — 92% ± 10.5% MFN2 cluster co-localization with BAX foci (n=847 cells). MFN2-at-apoptotic-fission-sites is Karbowski 2002 — drafting gap on mitofusins.md flagged.
• Beere 2000 cell line + Hsp70AAAA negative result (hsp70 verifier): wiki listed “S49 cells” — paper uses Jurkat T cell cytosolic extracts + 293T + MCF-7 + THP.1 + recombinant reconstitution. Hsp70AAAA (EEVD-motif alanine substitution) negative-result mechanism added. UniProt PTM features confirmed (Lys77 acetyl, Lys561 trimethyl by METTL21A, Ser631/Ser633 phospho).
• DOI corrections caught at seed time (no propagation needed):
  • de Grey 2002 SENS: brief 10.1196/annals.959.054 → 10.1111/j.1749-6632.2002.tb02115.x (apoptosenes seeder)
  • Mayer & Bukau 2005: brief 4464-4 → 4464-6 (hsp70 seeder; matches verified hsc70.md)
  • Akerfelt 2010: brief nrm2918 (Taipale HSP90 review) → nrm2938 (hsp70 seeder)
  • Ko 2011: brief 1112744108 → 1112317108 (tet2 seeder)
  • Karbowski 2002: brief jcb.200204140 (Actin-axon paper) → jcb.200209124 (mitochondrial-dynamics seeder)
  • 2 BUG-2 DOIs DROPPED (il-1b seeder): 10.1016/0092-8674(95)90420-4 (Zheng 1995 Il1b-/-, was Gerace paper) — verifier later resolved with correct Zheng 1995 Immunity DOI 10.1016/1074-7613(95)90154-x; 10.1038/342324a0 (Black 1989) — was 404, no replacement found
  • 6 of 12 lamp2 DOIs corrected (Cuervo&Dice 2000, Massey 2006, Bandyopadhyay 2008, Schneider 2014, Nishino 2000, Fujiwara 2013, Anguiano 2013, Bourdenx 2021)
  • apoptosenes seeder flagged He 2020 DOI: brief 10.1126/scitranslmed.aax5430 → archive has 10.1038/s41467-020-15838-0 (Nat Commun PROTAC paper); apoptosenes used the correct DOI per verified navitoclax.md

Propagation pass (this session)

• molecules/proteins/sirt1.md line 63 + line 193: AMPK PGC-1α Thr177/Ser538 unsourced resolved via cross-link to pgc-1alpha.md verified.
• pathways/ampk.md line 105: PGC-1α PTM site mouse-vs-human numbering note added with Jager 2007 attribution.
• pathways/ep300.md lines 44-45 (R12 carryover): ZZ + TAZ2 domain coords corrected per UniProt Q09472.
• hallmarks/dysbiosis.md lines 52, 84, 178: scfa-signaling references updated from “stub” to “verified R13”.
• Cross-page checks performed (no edits required): clonal-hematopoiesis.md (no 50-60% CMML claim), hematopoietic-stem-cells.md (no 50%-malignancy claim), epigenetic-alterations.md (TET cascade attribution generic, not stale-specific), chaperone-mediated-autophagy.md (Zhang 2008 strain claims already correct on that verified-partial page), alpha-synuclein.md (Cuervo 2004 framing already correct), nlrp3-inflammasome.md (Hoffman 2001 mutations already correct), alzheimers-disease.md (NLRP3/IL-1β framing already qualified).

ROADMAP updates

• Coverage summary refreshed (R12 → R13 state; ~230 pages).
• Implicit-stub queue refreshed; 8 R13 entries flipped [ ] → [x] with provenance pointers.
• Round 13 section added under Suggested seeding order with 8 R13 entries.
• Round 14+ candidates identified from refreshed lint queue: tet1, dnmt3a, asxl1 (CHIP driver triad completion), neural-stem-cells (13 inbound, deferred from R13), oligodendrocytes/astrocytes (cell types, 9 each), apoptosis-related study pages, hypotheses/mitohormesis (now strongly cross-linked via PGC-1α), hypotheses/antagonistic-pleiotropy.

Schema escalations (still pending user decision)

• CLAUDE.md update for type: framework — confirm framework pages do NOT carry verified fields (apoptosenes verifier removed them this round). Match the convention of sens-damage-categories and hallmarks-of-aging.
• CLAUDE.md update for retracted papers — formalize the bold “RETRACTED” prefix + retraction DOI convention used by pgc-1alpha seeder.
• R12 still pending: type: microbe — formalize for akkermansia-muciniphila.
• R12 still pending: biologic compound flag — biologic: true + who-inn: for canakinumab and future mAbs.

Outstanding TODOs surfaced (R14+ work)

• Implicit stubs created or referenced this round (not yet seeded): tet1, tet3, dnmt3a, asxl1, idax-cxxc4, idh1, idh2, 5hmc, caspase-1, asc, gsdmd, il-1r1, il-1ra, il-1a, anakinra, colchicine, pyroptosis, nrf1, tfam, gcn5, mitohormesis / mitohormesis, exercise, resveratrol, gpr41, gpr43, gpr109a, hdac, foxp3, regulatory-t-cells, gut-barrier, gut-microbiome, dietary-fiber, heat-shock-response, hsf1, dnaj, bag, hsp90, mff, fis1, mid49, mid51, oma1, yme1l, cathepsin-l, cathepsin-d, 5-alpha-reductase (R11 carryover).
• HSC verified-scope mildly stale (R12 carryover).
• mitofusins.md (drafted) drafting gap: Karbowski 2002 MFN2-at-apoptotic-fission-sites finding (92% co-localization, n=847) should be incorporated when mitofusins is verified.

Pages verified: 1 (partial scope — 5/8 primary sources verified against local PDFs; Ko 2011 and Fuster 2017 are no-fulltext-access; Ko 2013 IDAX download failed)

Sources checked and status:

• Ito 2010 (10.1038/nature09303) — VERIFIED against local PDF (pre-existing)
• Delhommeau 2009 (10.1056/NEJMoa0810069) — VERIFIED (downloaded this session)
• Moran-Crusio 2011 (10.1016/j.ccr.2011.06.001) — VERIFIED (downloaded this session)
• Cimmino 2017 (10.1016/j.cell.2017.07.032) — VERIFIED (downloaded this session)
• Agathocleous 2017 (10.1038/nature23876) — VERIFIED (downloaded this session)
• Ko 2011 (10.1073/pnas.1112317108) — NOT VERIFIED (not_oa / closed-access); DOI corrected via Crossref
• Fuster 2017 (10.1126/science.aag1381) — NOT VERIFIED (#gap/no-fulltext-access, download failed)
• Ko 2013 IDAX (10.1038/nature12052) — NOT VERIFIED (download failed, green OA but URL filter blocked)
• Tahiliani 2009 (10.1126/science.1170116) — NOT VERIFIED (download failed)
• Svensson 2022 — cross-referenced from verified canakinumab.md (R12); not re-verified here
• UniProt Q6N021 identity fields — spot-checked against live API; all confirmed

Corrections made (9):

1. Disease frequency table entirely wrong for Delhommeau 2009: Wiki had CMML ~50–60%, MDS ~25–35%, AML ~10–15%, MPN ~10–15% attributed to Delhommeau 2009. Actual Delhommeau 2009 figures: MDS 19% (15/81), MPN 12% (24/198), secondary AML 24% (5/21), CMML 22% (2/9), overall 15% (46/309). The higher CMML/MDS frequencies (~50%, ~30%) come from later literature (e.g., cited in Cimmino 2017); the incorrect attribution to Delhommeau 2009 has been corrected and a note added.
2. “~50% develop myeloid malignancy by 12 months” (Moran-Crusio 2011) — removed: Moran-Crusio 2011 reports CMML-like disease by 20 weeks as the primary endpoint; the 12-month/50% malignancy figure is not in this paper. Corrected to: “CMML-like disease by 20 weeks (splenomegaly >250 mg, monocytosis, myeloid dysplasia).”
3. Mouse model strain clarified (Moran-Crusio 2011): “Tet2 KO” → conditional Vav-Cre × Tet2-floxed on C57BL/6 × 129/SvEv hybrid background; n=15 per genotype for competitive transplants.
4. Ko 2013 IDAX DOI corrected (critical): DOI 10.1016/j.molcel.2013.05.004 resolves to an unrelated BMP-Smad paper (Xu et al. 2013, “Arginine Methylation Initiates BMP-Induced Smad Signaling”). Correct DOI is 10.1038/nature12052 (Ko M et al., Nature 2013, “Modulation of TET2 expression and 5-methylcytosine oxidation by the CXXC domain protein IDAX”). Corrected in footnote; claims from [^ko2013] remain unverified as download failed.
5. IDH/2HG citation misattribution corrected: Delhommeau 2009 was cited for the IDH/2HG/TET2 mutual exclusivity connection. Delhommeau 2009 (May 2009) predates IDH mutation characterization in myeloid cancer. Citation removed; paragraph tagged unsourced with note to cite Dang et al. 2009 / Xu et al. 2011.
6. Ito 2010 scope clarified: Primary paper focus is Tet1’s role in ES-cell self-renewal and ICM specification. TET2/TET3 5mC→5hmC activity is the secondary finding. In-vitro work used baculovirus-SF9-expressed catalytic domains; in-vivo used U2OS/HEK293 overexpression. Footnote updated.
7. Cimmino 2017 description corrected: Wiki said “ascorbate restored TET2 function in TET2-deficient leukemia cell lines and mouse models.” Corrected to: primary model is a reversible RNAi Tet2-knockdown mouse (VTA-shTet2); vitamin C mimics Tet2 restoration, not “restores”; effect quantified (2-fold 5hmC increase, ~30% 5mC decrease, blocks re-plating beyond P5, suppresses AML PDX colony formation); PARP inhibitor synergy noted.
8. Agathocleous 2017 description corrected: Wiki said “plasma ascorbate levels are markedly lower in leukemia patients vs healthy controls” (that finding is not the main result). Corrected to: HSCs/MPPs accumulate ascorbate 2–20-fold above restricted progenitors; ascorbate depletion in Gulo-/- mice increases HSC frequency partly via Tet2; cooperation with Flt3-ITD; compound Gulo-/-;Tet2Δ/Δ mice show attenuated effect.
9. Steps 2–3 (5fC, 5caC) attribution corrected: Wiki attributed the full iterative cascade to Tahiliani 2009 and Ito 2010. Those papers only establish step 1 (5mC→5hmC). Steps 2–3 were established by subsequent 2011 papers; tagged unsourced pending proper citation.

Unverifiable claims:

• Ko 2011 HSC self-renewal and differentiation data — closed-access
• Fuster 2017 quantitative plaque area, exact IL-1β/IL-6 fold-changes — no-fulltext-access
• Ko 2013 IDAX-TET2 targeting mechanism — download failed; claims tagged

Downstream propagation needed (for main agent):

• clonal-hematopoiesis.md — check if it carries the wrong CMML/MDS frequencies attributed to Delhommeau 2009; if so, correct or add note that Delhommeau 2009 figures are 15% overall, not 50% CMML
• atherosclerosis.md — uses Fuster 2017; no new corrections from this pass (still no-fulltext-access)
• hematopoietic-stem-cells.md — check “50% malignancy by 12 months” language if inherited from old tet2.md claim
• epigenetic-alterations.md — check if 5fC/5caC steps are properly attributed (not to Tahiliani/Ito 2010)
• Studies pages for Delhommeau 2009 and Moran-Crusio 2011 do not yet exist; when created, use the verified figures from this pass

[2026-05-05] verify | pathways/scfa-signaling.md

Pages verified: 1 (partial scope — 7/9 primary sources verified against PDF; 2 are not_oa)

Sources checked and status:

• Claesson 2012 (10.1038/nature11319) — VERIFIED (local PDF); 3 corrections
• Kimura 2011 (10.1073/pnas.1016088108) — VERIFIED (downloaded); 2 corrections
• Singh 2014 (10.1016/j.immuni.2013.12.007) — VERIFIED (downloaded); strain detail added + Treg quantification added
• Smith 2013 (10.1126/science.1241165) — VERIFIED (downloaded); 2 corrections
• Arpaia 2013 (10.1038/nature12726) — VERIFIED (downloaded); mechanistic nuance corrected and expanded
• Gao 2009 (10.2337/db08-1637) — VERIFIED (downloaded); all claims confirmed; no corrections
• Donohoe 2011 (10.1016/j.cmet.2011.02.018) — VERIFIED (downloaded); 1 major correction
• Maslowski 2009 (10.1038/nature08530) — NOT VERIFIED (not_oa); tagged no-fulltext-access in footnote
• Furusawa 2013 (10.1038/nature12721) — NOT VERIFIED (not_oa); tagged no-fulltext-access in footnote

Corrections made:

• Donohoe 2011 — “60–70% energy from butyrate” removed from attribution: The percentage does not appear in Donohoe 2011. Paper shows butyrate is PRIMARY energy source via etomoxir rescue experiment; GF colonocytes have 70% drop in mitochondrial NADH/NAD+; butyrate ex vivo rescues oxidative phosphorylation from ~30% to ~70% of CONV-R levels. The “60–70%” figure is from Roediger 1980 (Gut); noted and unsourced applied.
• Kimura 2011 — phenotype description corrected: “reduced body fat and altered energy expenditure” → primary finding is reduced resting heart rate (682 ± 10 vs. 610 ± 16 bpm, n=12/group), reduced SNS outflow, lower body temperature, reduced UCP-1 and oxygen consumption in Gpr41-/- mice. Signaling mechanism corrected to Gβγ-PLCβ-MAPK. Adipose Gpr41 expression noted as not confirmed in this study (contradicts prior reports).
• Smith 2013 — delivery route corrected: “SCFA gavage” → SCFAs provided in drinking water at 150 mM. Propionate (not “propionate and butyrate equally”) is the primary Ffar2/GPR43 ligand; HDAC6/9 downregulation is Ffar2-dependent and downstream, not an independent parallel mechanism.
• Singh 2014 — Treg quantification and strain detail added: ~40% reduction in colonic Foxp3+ Tregs in Niacr1-/- mice. Main colitis experiments use Swiss Webster mice (not C57BL/6); C57BL/6 background only for Il18-/- and ApcMin/+ crosses.
• Claesson 2012 — Faecalibacterium attribution corrected: Main text names Coprococcus and Roseburia as depleted in long-term care; Faecalibacterium is supplementary context only. Cohort strata added (n: community 83, day-hospital 20, rehabilitation 15, long-term care 60); mean age 78 ± 8; metagenome dataset noted (125.9 Gb, 27/29 subjects); inflammatory marker differences (CRP, IL-6) added.
• 2013 Treg trifecta mechanistic reconciliation expanded: Arpaia 2013 effect is CNS1-dependent and GPR-independent (Gpr109a-/- DCs non-blocking; pertussis toxin non-blocking); butyrate drives colonic Treg only via local delivery; propionate drives peripheral but not colonic Treg; acetate inactive. Smith 2013 GPR43/HDAC relationship clarified as sequential not alternative.
• SCFA luminal concentration: Added Smith 2013 cited range (50–100 mM) alongside wiki’s “80–150 mM”; identified Cummings 1987 (Gut) as canonical primary source for molar ratio — not yet on this page, flagged unsourced.
• Auto-extraction banner removed upon verification flip.

Unverifiable claims (closed-access):

• Maslowski 2009 (Gpr43-/- colitis/arthritis, GPR43 required for SCFA anti-inflammation) — no-fulltext-access
• Furusawa 2013 (HDAC3-specific mechanism, GPR43-independence) — no-fulltext-access

Downstream propagation needed (for main agent):

• dysbiosis.md (verified) — Claesson 2012 cited there; check consistency of n=178+13 description, Coprococcus/Roseburia naming, frailty correlation. Add cross-link to scfa-signaling as canonical home for the SCFA mechanism.
• akkermansia-muciniphila.md (verified-partial R12) — propionate + acetate claim consistent; no correction needed.
• molecules/proteins/gpr43.md, gpr41.md, gpr109a.md — when seeded, inherit corrected phenotype descriptions (especially Kimura 2011 GPR41 phenotype and ~40% Treg reduction via GPR109A from Singh 2014).

[2026-05-05] verify | molecules/proteins/lamp2.md

Pages verified: 1 (partial — 3 of 12 cited sources verified from local PDF; 3 confirmed via Crossref/Europe PMC abstract; remainder closed-access or PDF-pending)

Sources checked:

• Bandyopadhyay 2008 (10.1128/MCB.02070-07) — VERIFIED against local PDF
• Zhang & Cuervo 2008 (10.1038/nm.1851) — VERIFIED against local PDF
• Massey 2006 (10.1073/pnas.0507436103) — VERIFIED against local PDF (downloaded 2026-05-05)
• Cuervo & Dice 2000 (10.1074/jbc.M002102200) — Europe PMC abstract + Crossref; local PDF failed
• Nishino 2000 (10.1038/35022604) — DOI confirmed via Crossref; not_oa
• Schneider 2014 (10.1016/j.cmet.2014.06.009) — DOI/authorship confirmed; local PDF corrupted (BUG-5)

Corrections made:

• Zhang 2008 strain: “C57BL/6 transgenic mice (TF-LAMP2A)” → “FVB transgenic mice (Alb-Tet-off-L2A)”
• Zhang 2008 age: “22–24 month” → “22–26 month”
• Zhang 2008 phenotypes: “albumin secretion, glucose homeostasis” removed; not directly measured in paper
• Cuervo 2000 age: “24-month” → “22-month” rat liver
• Cuervo 2000: added that paper includes human fibroblast data (not “rat liver only”)
• Bandyopadhyay 2008 complex characterization: “by BN-PAGE” → “by four independent methods (native gel, BNE, gel filtration, sucrose density gradient)”
• ~36h LAMP-2A half-life claim removed from step 7 of CMA mechanism — not stated in Bandyopadhyay 2008; tagged unsourced
• Massey 2006 attribution: clarified this paper demonstrates knockdown (not overexpression); overexpression result from prior Cuervo & Dice papers
• Nishino 2000 DOI disambiguation: 35022604 (human disease) vs. 35022595 (mouse KO) noted in footnote

Downstream propagation needed (for main agent):

• chaperone-mediated-autophagy.md — check strain claims from Zhang 2008 (C57BL/6 → FVB)
• hsc70.md — check LAMP-2A complex characterization method language
• alpha-synuclein.md — consistent; no correction

[2026-05-05] verify | molecules/proteins/pgc-1alpha.md

Pages verified: 1 (partial — Leick 2010 closed-access; canonical-DB identity fields not re-checked on this pass)

Sources checked against local PDFs:

• Puigserver 1998 (10.1016/s0092-8674(00)81410-5) — VERIFIED
• Rodgers 2005 (10.1038/nature03354) — VERIFIED; acetylation site list corrected (mouse vs human numbering)
• Lerin 2006 (10.1016/j.cmet.2006.04.013) — VERIFIED; downloaded and read
• Jager 2007 (10.1073/pnas.0705070104) — VERIFIED; downloaded and read; Thr177/Ser538 confirmed as mouse sequence
• Li 2007 (10.1038/nature05861) — VERIFIED; Akt site corrected S265+S568 → S570 primary site
• Hubbard 2013 (10.1126/science.1231097) — VERIFIED; K778 confirmed as mouse numbering throughout
• Lin 2004 (10.1016/j.cell.2004.09.013) — VERIFIED; downloaded and read
• Ristow 2009 (10.1073/pnas.0903485106) — VERIFIED; downloaded and read; n and design clarified
• Leick 2010 (10.1016/j.exger.2010.01.011) — UNVERIFIABLE (not_oa, closed-access)
• Wenz 2009 retraction (10.1073/pnas.1619713114) — confirmed via Crossref; retraction December 2016

Corrections made:

• GCN5 acetylation site table restructured: was single-column UniProt human numbers; now two-column showing mouse source numbering (K77/K144/K183/K253/K270/K277/K320/K346/K412/K441/K450/K757/K778 per Rodgers 2005) alongside human UniProt equivalents (+2 offset). K757 (mouse) = K758 (human) added as a 12th GCN5 site; K778 (mouse) = K779 (human) is the SIRT1 site.
• Frontmatter PTMs corrected: K758→K778 (mouse) for SIRT1 entry; S568→S570 for Akt entry; removed S265 as Akt site (not in Li 2007); added sourcing notes to each PTM entry.
• Akt phosphorylation claim corrected: “S265 + S568” → “Ser570 as primary site” per Li 2007. S265 is unsourced from Li 2007 and tagged unsourced.
• Lin 2004 neurodegeneration corrected: “aged Pgc-1α-null mice” → “3-month-old mice” (striatal lesions found at 3 months in Lin 2004, not in aged mice).
• Lin 2004 cardiac claim removed: “cardiac dysfunction: moderate cardiac failure under stress” was not reported in Lin 2004; removed.
• Lin 2004 phenotype quantified: added specific numbers from paper (body temp drops to 33.5°C, UCP1 induction 45% of WT, D2 mRNA -50%, mitochondrial gene expression -30 to -60%, +40% movement).
• Domain table corrected: “S265, S568” → “S570 per Li 2007” in the ~400–600 region row.
• Hubbard 2013 K778/K779 clarified: explicit note that K778 is mouse sequence, K779 is human UniProt; both refer to the same residue.
• Ristow 2009 footnote updated: n=40 correctly described with two-part design (open-label n=16 + double-blind n=24); NCT00638560 added; specific antioxidant doses added; status updated to “locally available in archive.”
• Lin 2004 footnote updated: corrected phenotype description; “not downloaded (pending)” → “locally available in archive.”
• Jager 2007 footnote updated: expanded with mouse-vs-human numbering note; “not downloaded (pending)” → “locally available in archive.”
• Lerin 2006 footnote updated: fully detailed; “not downloaded (pending)” → “locally available in archive.”
• Leick 2010 footnote updated: “not OA” → “#gap/no-fulltext-access”; claim in aging section tagged accordingly.

Unverifiable (closed-access):

• Leick 2010 (10.1016/j.exger.2010.01.011) — tagged no-fulltext-access; aging-muscle claim remains unverified

Schema gap noted:

• No formal convention in CLAUDE.md for documenting retracted papers. The seeder’s approach (bold “RETRACTED” prefix in footnote with retraction DOI) is appropriate. Recommend user formally adopt this in CLAUDE.md under Citation discipline.

Downstream propagation needed (for main agent):

• sirt1.md — Jager 2007 was tagged unsourced for AMPK Thr177/Ser538 site on sirt1.md (from prior verification pass); pgc-1alpha.md now has the verified Jager 2007 citation. Main agent should update sirt1.md to resolve that gap.
• ampk.md — AMPK→PGC-1α phosphorylation sites (Thr177/Ser538 mouse / T178/S539 human) now verified; check ampk.md for any stale site numbers.
• cbp-p300.md — attribution caveat (GCN5 is writer, not CBP/p300) is consistent with pgc-1alpha.md; no corrections needed but cross-link should be verified on next pass.
• mitochondrial-biogenesis.md — may have PGC-1α claims that reference Lin 2004 phenotypes or Wenz 2009 lifespan; check and correct if so.

[2026-05-05] verify | molecules/proteins/il-1b.md

Pages verified: 1 (partial — 3 of 11 primary sources verified against PDF; remainder closed-access/pending)

Sources checked:

• Heneka 2013 (10.1038/nature11729) — VERIFIED against local PDF
• Hoffman 2001 (10.1038/ng756) — VERIFIED against local PDF; 2 factual corrections
• Ridker 2017 NEJM (10.1056/NEJMoa1707914) — VERIFIED against local PDF; consistent with canakinumab.md
• UniProt P01584 — chain/cleavage features verified via REST API (propeptide 1–116, mature chain 117–269, SpeB sites 105–106 and 115–116 all confirmed)
• Zheng 1995 (10.1016/1074-7613(95)90154-x) and Horai 1998 (10.1084/jem.187.9.1463) — identified via PubMed + Crossref as canonical Il1b-/- KO sources; DOIs confirmed

Corrections made:

• Hoffman 2001 — mutation names corrected: “L353P / R260W type mutations” → actual mutations from paper: A439V, V198M, E627G (FCAS) and A352V (MWS), all exon 3 of CIAS1. Original names not present in this paper.
• Hoffman 2001 — n families corrected: “~10 families” → “3 FCAS families + 1 MWS family (4 families total); >100 normal controls.” Paper studied exactly 4 families.
• Heneka 2013 — AD/microglia claim qualified: “IL-1β is a driver of microglia M1 polarisation” → clarified that Heneka 2013 implicates the full NLRP3/caspase-1 axis (not IL-1β alone) in M1→M2 skewing; added specific endpoints (70% FA-Aβ reduction, MWM rescue, 16-month age, M2 marker upregulation).
• Heneka 2013 footnote — n corrected: “n=approx 10–20/group” → “n not stated per group in main text” (n not given in 5-page body); specific quantitative endpoints added.
• Hoffman 2001 footnote — corrected: n and mutation list updated to match paper.
• Il1b-/- KO gap resolved: unsourced removed; Zheng 1995 (Immunity) added as canonical single Il1b-/- KO; Horai 1998 (J Exp Med) added as multi-KO paper; two new footnotes [^zheng1995] and [^horai1998] added.

Unverifiable (closed-access / pending):

• Auron 1984, March 1985, Cerretti 1992, Black 1989, Howard 1991, Franceschi 2000, Shi 2015, Fuster 2017 — tagged no-fulltext-access (partial) in Limitations section.

Downstream propagation needed (for main agent):

• nlrp3-inflammasome.md — should link il-1b as canonical IL-1β home; check whether Hoffman mutation names are duplicated there and correct if so.
• alzheimers-disease.md — check for “IL-1β drives M1 microglia” claim; add NLRP3/caspase-1 axis qualifier if present, consistent with Heneka 2013.

[2026-05-05] verify | molecules/proteins/hsp70.md

Pages verified: 1 (partial — 1 of 6 sources OA/local)

• molecules/proteins/hsp70.md — corrections:
  • Beere 2000 footnote: removed erroneous “murine lymphoma S49 cells” (S49 not used experimentally in this paper); replaced with accurate model list: Jurkat T cell cytosolic extracts, 293T/MCF-7 overexpression cells, THP.1 fractionation, recombinant reconstitution
  • Beere 2000 footnote: expanded mechanism note to capture key negative finding — Hsp70AAAA mutant (EEVD-motif alanine substitution) lacks inhibitory activity; Hsp70 does not block Apaf-1 oligomerization
  • Locke 1996 body text: “47% reduction in HSF1 activation” → “47% reduction in HSF1 DNA-binding activity” (precise measurement term per footnote); added no-fulltext-access tag
  • Heydari 1993 body text: added “in liver” (tissue specificity, consistent with footnote) to prevent ambiguity; added no-fulltext-access tag
  • Hsu 2003 body text: added no-fulltext-access tag on lifespan extension claim
  • Verified flag flipped to true with partial scope

Pages unverifiable (closed-access): Heydari 1993, Locke 1996, Hsu 2003, Akerfelt 2010 — all tagged no-fulltext-access

UniProt verified (live API): P0DMV8 (HSPA1A, 641 aa), P0DMV9 (HSPA1B, 641 aa), P34931 (HSPA1L, 641 aa) — all match table; PTM sites (Ala-2 N-acetyl, Lys-77 acetyl, Lys-561 trimethyl, Ser-631/Ser-633/Thr-636 phospho) confirmed

Downstream pages to check (main agent):

• molecules/proteins/apaf-1.md — cites HSP70-Apaf-1 interaction; apaf-1.md footnote does not inherit the S49 error (its own verified entry is correct)
• processes/chaperone-mediated-autophagy.md — CASA/BAG3 claims on hsp70.md are consistent with that page; no propagation needed unless CASA page is re-verified

[2026-05-05] verify | frameworks/apoptosenes.md

Pages verified: 1

• frameworks/apoptosenes.md — framework MOC (navigational overlay); no primary-source quantitative claims originated here

Corrections made:

• SCAP table HUVEC row: [[bcl-xl]], [[bcl-2]], EFNB1/EFNB3 → [[bcl-xl]], EFNB1/EFNB3. BCL-2 is not a HUVEC-dominant SCAP node per verified atomic pages (bcl-2.md, bcl-xl.md, quercetin.md, senolytics.md all consistent: BCL-2 is an IMR90 fibroblast SCAP node; HUVECs rely primarily on BCL-xL + EFNB1).
• Body text line 126: [[cellular-senescence]] process → [[hallmarks/cellular-senescence]] (no separate processes/cellular-senescence.md exists; bare link would have resolved but the label “(process page)” was inaccurate).
• See also section: Replaced [[cellular-senescence]] (process page) with a note clarifying processes/cellular-senescence.md is not yet seeded and redirecting to [[hallmarks/cellular-senescence]].
• Schema/frontmatter: Removed verified: false, verified-date: null, verified-by: null fields — CLAUDE.md schema for type: framework does not include verified fields, and neither sens-damage-categories.md nor hallmarks-of-aging.md carry them. Framework pages are navigational overlays; the verified discipline applies to atomic pages that originate quantitative claims.
• ⚠️ banner removed: Framework framing is structurally sound; SCAP data deferred to verified atomic pages as intended.

DOI spot-check: de Grey 2002 10.1111/j.1749-6632.2002.tb02115.x confirmed via Crossref (title “Time to Talk SENS: Critiquing the Immutability of Human Aging”, Annals NYAS, 2002). All other DOIs (Zhu 2015, 2016; Yosef 2016; Baar 2017; He 2020; Baker 2011, 2016; Yousefzadeh 2018) confirmed via Crossref — all correct.

Wikilink audit: all protein pages (bcl-2, bcl-xl, bcl-w, mcl-1, bax, bak, foxo4, p53), compound pages (navitoclax, a1331852, dasatinib, quercetin, fisetin), pathway pages (bcl-2-family-signaling, pi3k-akt-pathway, nf-kb, p53-pathway, apoptosis-pathway), process pages (sasp), hallmark pages (cellular-senescence, etc.), framework pages (sens-damage-categories, hallmarks-of-aging), and intervention page (senolytics) all resolve to existing files. Study pages referenced (de-grey-2002-sens, zhu-2015-scap-senolytics, etc.) are implicit stubs — none exist except yousefzadeh-2018-fisetin-senolytic; consistent with pattern across wiki.

Schema decision (for main agent): framework pages should NOT carry verified fields going forward. The two pre-existing framework pages confirm this. The seeder erred in adding those fields to apoptosenes.md; they have been removed.

Downstream propagation needed: none — this is a framework MOC with no originating quantitative claims; the BCL-2/HUVEC SCAP correction is already correct on all atomic pages (bcl-2.md, bcl-xl.md, quercetin.md, senolytics.md).

[2026-05-05] ingest | scfa-signaling (pathway)

• added: pathways/scfa-signaling.md
• canonical IDs: Reactome R-HSA-444209 (“Free fatty acid receptors”) confirmed; KEGG null (no dedicated entry); Reactome R-HSA-211897 confirmed incorrect (Cytochrome P450 by substrate type — not SCFA)
• DOIs cited (9): all confirmed via Crossref + DOI lookup; Claesson 2012 has local PDF
• gaps surfaced: SCFA molar ratio/concentration claims need primary-source citation; dose-response in humans unclear; SCFA contribution to inflammaging magnitude unresolved; GPR vs. HDAC mechanistic split in Treg induction not reconciled
• implicit stubs created: gpr41, gpr43, gpr109a, hdac, foxp3, gut-microbiome, dietary-fiber, regulatory-t-cells, gut-barrier
• schema note: hallmarks: ["[[dysbiosis]]",...] — dysbiosis used as a hallmark-level category in frontmatter, consistent with R10a usage; if hallmarks ontology is tightened, this field may need updating

[2026-05-05] Round 12 batch summary (high-leverage post-lint adds + R10e/R10f close-out)

Seeders (6 in parallel, all completed)

After R11 the post-lint implicit-stub queue surfaced these as the highest-priority unblocks:

• phenotype: phenotypes/clonal-hematopoiesis.md — 17 inbound (highest in queue)
• protein: molecules/proteins/drp1.md — 14 inbound (mitochondrial fission GTPase; companion to mitofusins)
• protein family: molecules/proteins/cbp-p300.md — 12 inbound (CREBBP+EP300 paralog-pair page)
• pathway: pathways/p16-rb-pathway.md — 9 inbound (senescence reinforcement arm)
• microbe (NEW SCHEMA): microbiome/akkermansia-muciniphila.md — 6 inbound (introduces type: microbe)
• compound: molecules/compounds/canakinumab.md (anti-IL-1β IgG1κ; CANTOS anchor)

Verifiers (6 in parallel, all completed)

Major cross-cutting corrections:

• Bouzid 2023 CHIP-AD direction INVERTED (caught by clonal-hematopoiesis verifier): wiki framing said CHIP “contributes to neuroinflammation and neurodegeneration”; Bouzid et al. 2023 (Nat Med, doi:10.1038/s41591-023-02397-2) actually shows CHIP is protective for AD (OR=0.64; P=3.8×10⁻⁵; MR-supported causal direction). Section rewritten on clonal-hematopoiesis.md. neurodegeneration.md already had correct framing — no propagation needed there. Cross-link to clonal-hematopoiesis.md recommended for future neurodegeneration verifier pass.
• Jaiswal 2014 CIs corrected (clonal-hematopoiesis verifier): CHD 95% CI 1.2-3.4 → 1.2-3.5; ischemic stroke 95% CI 1.4-4.8 → 1.3-4.8. Verifier propagated directly to cell-types/hematopoietic-stem-cells.md. HSC verified-scope field is now slightly stale on those CIs — flag for next lint pass.
• Jaiswal 2017 HR reframed (clonal-hematopoiesis verifier): “approximately 2× CHD” → HR=1.9 (95% CI 1.4-2.7; P<0.001) per Fig. 1A fixed-effects meta-analysis.
• Driver gene table rebuilt with raw variant counts (clonal-hematopoiesis verifier): unsourced percentages (DNMT3A ~46%, TET2 ~16%, ASXL1 ~11%) replaced with Jaiswal 2014 Fig. 2A counts (DNMT3A 403 / TET2 72 / ASXL1 62 / TP53 33 / JAK2 31 / SF3B1 27 / SRSF2 22). Genovese 2014 ASXL1>TET2 ranking discrepancy noted.
• Wakabayashi 2009 Cre driver corrected (drp1 verifier): brain-specific KO used En1-Cre (cerebellum/midbrain), NOT Nestin-Cre. Death at 36 hours postnatal, NOT 3 weeks. Phenotype is cerebellar developmental defect (Purkinje cell loss at P0), not progressive degeneration.
• Mdivi-1 mechanism CRITICAL re-quantification (drp1 verifier): wiki claimed “IC50 ~1-10 µM as DRP1 inhibitor”; Bordt 2017 actually shows Ki >1.2 mM for human DRP1 (the IC50 cited is for yeast Dnm1, not human DRP1). Effective concentration range corrected from 10-50 µM → 25-100 µM. Mdivi-1 is primarily a Complex I inhibitor at functional doses, not a clean DRP1 inhibitor — this caveat is now load-bearing on the page.
• Cribbs & Strack 2007 isoform numbering clarified (drp1 verifier): paper uses rat Drp1 splice variant 1; the residue is Ser656 in rat = Ser637 in human isoform 1. Isoform-numbering note added to PTM table and footnote.
• p53 K386 removed from acetylation site list (cbp-p300 verifier): Gu & Roeder 1997 explicitly states K386 was completely unacetylated. Wiki had K370/K372/K373/K381/K382/K386 → corrected to K370/K372/K373/K381/K382. The MDM2-displacement mechanism was also removed from the Gu 1997-sourced section (that mechanism is from later work, not in the 1997 paper).
• PGC-1α CBP/p300 writer attribution caveated (cbp-p300 verifier): Rodgers 2005 identifies SIRT1 as the eraser but never demonstrates CBP/p300 as the writer. Attribution is inferential — flagged in substrate table, body, and footnote on cbp-p300.md. Cross-checked sirt1.md — that page already correctly states “13 acetylation sites mapped by mass spectrometry in Rodgers 2005” without naming a specific writer; no propagation needed.
• EP300 domain coordinates corrected on ep300.md (cbp-p300 verifier flagged + main-agent propagation this pass): ZZ zinc-finger ~1723–1770 → 1665–1713 (UniProt Q09472); TAZ2 ~1834–1890 → 1728–1809 (UniProt Q09472). Also: Petrij 1995 patient cohort n=24 → n=16 RSTS patients examined by PTT (the 24 is not in the paper).
• Heintzman 2009 cell line scope corrected (cbp-p300 verifier): “9 cell types” → “5 human cell lines (HeLa, GM06690, K562, ES, BMP4-induced ES)”.
• Sharpless 2001 tumor spectrum and n’s added (p16-rb-pathway verifier): “fibrosarcomas and B-cell lymphomas” → actual Table 1 spectrum (soft-tissue sarcoma 5, splenic lymphoma 4, melanoma 1); n per group (26/40/39); P=0.008 log-rank. Added p19ARF as primary mediator of passage-induced MEF growth arrest in this paper (NOT p16).
• Krimpenfort 2001 melanoma claim CRITICAL correction (p16-rb-pathway verifier): the wiki claimed “Ink4a−/− developed melanoma with incomplete penetrance”; the paper actually shows simple Ink4a*/+ heterozygotes had NS spontaneous tumors (P=0.15, 2/12 lymphomas). Melanoma was only seen in compound Ink4a/Δ2,3 genotype with DMBA treatment* (7/14, P<0.00005). p16 heterozygosity alone is insufficient for spontaneous melanoma in this model.
• Baker 2016 SASP cytokine list corrected (p16-rb-pathway verifier): “IL-6, MMP3 and other targets” → IL-6, IL-1α, TNFα per Extended Data Fig. 3b. MMP3 is not the SASP factor Baker 2016 reports for the Cdkn2a-clearance experiment.
• Akkermansia n-value conflict resolved (akkermansia-muciniphila verifier): seeder used n=32 enrolled / 27 completed (from task brief); verifier confirmed against PDF that n=40 enrolled / 32 completed (placebo 11 / pasteurized 12 / live 9) is correct, matching the verified dysbiosis.md figures. Also: genome size 2.66 Mb → 2.664 Mb (2,664,102 bp exact); protein-coding ORFs ~2,200 → 2,176; body weight delta “−2.27±0.92 kg” with P=0.09 (NS) explicit; inclusion criteria corrected to NCEP ATP III metabolic syndrome (BMI >25 + ≥3/5 components + HOMA-IS <75% + HbA1c ≤7.5%); plasma LPS reduction added; γGT −24% (P=0.009) added; PERMANOVA R²=0.075 p=0.18 added.
• Canakinumab CANTOS precision (canakinumab verifier): Svensson 2022 characterization “post-hoc” → pre-specified exploratory genomic substudy; subgroup n’s specified (3,923 sequenced; 338 CHIP carriers; TET2 placebo 31 vs canakinumab 71); HR 0.38 (95% CI 0.15-0.96; P=0.04) for TET2-CHIP MACE benefit; P-interaction 0.14 (NS); paper itself notes 26% power. CANTOS infection signal sharpened to 0.31 vs 0.18 events/100 PY (P=0.02) with non-tilde precision. Ridker 2017 NEJM downloaded + Svensson 2022 downloaded this pass (Svensson was previously failed; bronze OA worked).
• DOI corrections caught at seed time (no verifier propagation needed):
  • Karanam 2006 brief DOI 10.1074/jbc.M510590200 → 10.1074/jbc.M608813200 (cbp-p300)
  • Lee & Finkel 2009 brief DOI 10.1016/j.cell.2009.05.027 → 10.1074/jbc.M807135200 (cbp-p300; cross-validated against ep300.md)
  • Roelfsema 2005 brief DOI 10.1086/426462 → 10.1086/429130 (cbp-p300; cross-validated against ep300.md)
  • Sharpless 2004 brief DOI 10.1038/sj.emboj.7600142 (Haemophilus paper) → 10.1038/sj.onc.1207074 (p16-rb-pathway)
  • Canakinumab brief ChEMBL CHEMBL1201827 (panitumumab) → CHEMBL1201834 (canakinumab — verified via API)

Propagation pass (this session)

• molecules/proteins/ep300.md lines 44-45: ZZ + TAZ2 domain coordinates corrected per UniProt Q09472 (per cbp-p300 verifier).
• hallmarks/dysbiosis.md lines 38, 140, 157, 177: [[akkermansia-muciniphila]] — stub → verified-partial, R12 (4 occurrences).
• cell-types/hematopoietic-stem-cells.md: CHD + stroke CIs already updated by clonal-hematopoiesis verifier; verified-scope text mildly stale.
• Cross-page checks performed (no edits required): chronic-inflammation.md (CANTOS data already precise); atherosclerosis.md (CANTOS data already precise); nf-kb.md (IL-1β-only framing correct); senomorphics.md (canakinumab vs anakinra IL-1α/β distinction correct); sirt1.md (Rodgers 2005 PGC-1α attribution correct); p53.md and p53-pathway.md (no K386 or MDM2-displacement to correct).

ROADMAP updates

• Coverage summary refreshed (R11→R12 state; ~222 pages; 81%+ verified).
• Stale [ ] entries marked [x] with provenance pointers (oxphos, ubiquitin-proteasome-system, grb2, skeletal-muscle, myocardium, nlrp3-inflammasome, age-1, free-radical-theory-of-aging — all previously seeded but not flipped).
• R10e formally CLOSED — alias resolution decision documented (Obsidian alias-blind lint script will continue false-positives for inflammaging/irs1/bnip3l/bcl2-family-pathway/mfn1/mfn2; manual filter on output until lint script learns frontmatter aliases).
• R10f formally CLOSED — study page expansion deferred indefinitely (atomic-entity footnotes carry quality signals; dedicated study pages duplicate without unlock).
• Round 12 section added under Suggested seeding order with 6 R12 entries.
• Round 13+ candidates ranked from current implicit-stub queue (lamp-2a, hsp70, mitf, neural-stem-cells, oligodendrocytes, astrocytes, fibroadipogenic-progenitors, mitohormesis, antagonistic-pleiotropy, elamipretide).

Outstanding TODOs surfaced

• Schema escalation: type: microbe — akkermansia-muciniphila.md introduces this new entity type. CLAUDE.md should be updated to formalize the frontmatter shape if microbiome pages expand. Suggested required fields: ncbi-taxonomy, phylum, family, gram-stain, oxygen-tolerance, host, hallmarks, verified. Optional: genome-size-mb, key-strains, discovered. Decision deferred to user.
• Schema escalation: biologic compound pages — canakinumab.md notes that CLAUDE.md compound frontmatter assumes small molecules. For mAbs, pubchem-cid: null is necessary; consider adding optional biologic: true flag and who-inn: field if more biologic compounds are anticipated. Decision deferred to user.
• Implicit stubs created (not yet seeded): tet2, dnmt3a, asxl1, il-1b, il-6, mid49, mid51, mff, fis1, mitochondrial-dynamics, p16, rb, cdk4, cdk6, cyclin-d1, e2f1, p14arf, palbociclib, gut-microbiome, scfa-signaling, tlr2, gut-barrier, amuc1100, berberine, anakinra, colchicine, pgc-1alpha, histone-acetylation.
• Closed-access papers needing institutional access: Smirnova 2001 Mol Biol Cell, Otera 2010 J Cell Biol, Friedman 2011 Science, Waterham 2007 NEJM (for drp1 verification); Quelle 1995, Serrano 1993, Hara 1996, Alcorta 1996, Ruas&Peters 1998, Sharpless 2004 Oncogene (p16-rb-pathway); Ridker 2017 Lancet (canakinumab); Derrien 2004, Plovier 2017, Png 2010 (akkermansia); Thompson 2004, Brunet 2004 (cbp-p300); Cho 2009 (drp1 — green OA, download failed).
• Feature request candidate: Lint script alias-blindness — currently can’t resolve [[inflammaging]] to chronic-inflammation.md despite frontmatter alias. Adding alias-resolution would reduce false-positive broken-link reports by ~30 entries.

pathways/p16-rb-pathway.md

• page: pathways/p16-rb-pathway.md
• sources verified against full PDF (5): Krishnamurthy 2004, Sharpless 2001, Krimpenfort 2001, Baker 2016, Tao 2017
• sources unverifiable: Quelle 1995, Serrano 1993, Hara 1996, Alcorta 1996, Ruas & Peters 1998, Sharpless 2004 Oncogene (all not_oa); Baker 2011 (OA green but download failed — 0 candidate URLs)
• corrections made (8):
  1. Krishnamurthy 2004 age range: “1 versus 8–26 months” → “2.5 months (young) vs 26 months (old)” per Methods
  2. Krishnamurthy 2004 tissue list: “kidney, ovary, heart, lymphocytes” → uterus (>96x), cecum (>70x), kidney (>34x), ovary (30x) as highest; hematopoietic compartment data added from Fig. 2B; geometric mean ~9.7-fold added; CR attenuation pattern added
  3. Krishnamurthy 2004 quantification: Arf ~3.5-fold geometric mean comparator added
  4. Sharpless 2001 tumor spectrum: “fibrosarcomas and B-cell lymphomas” → soft-tissue sarcoma (5), splenic lymphoma (4), melanoma (1); n per group (26/40/39) and P=0.008 log-rank added; p19ARF role in MEF senescence added
  5. Krimpenfort 2001: “Ink4a−/− developed melanoma and fibrosarcoma with incomplete penetrance” → corrected — simple Ink4a*/+ heterozygotes NS (P=0.15, 2/12); melanoma requires compound Ink4a*/Δ2,3 genotype + DMBA (7/14, P<0.00005); spontaneous vs carcinogen-induced distinguished
  6. Baker 2016 SASP cytokines: “IL-6, MMP3 and other targets” → “IL-6, IL-1α, and TNFα” per Extended Data Fig. 3b
  7. Baker 2016 lifespan wording: “~24–27% mixed; ~24–25% B6” → “27% (mixed) / 24% (B6)” precise; range 17–35% across sex subgroups confirmed
  8. Tao 2017: drug name “ribociclib/LEE011” → “LEE011 (ribociclib)”; SAHF claim removed from body (paper does not demonstrate SAHF); cell-line context added (7/8 G1 arrest; THP-1 exception)
• verified: true (partial scope — 5/13 cited sources have local PDFs; remainder tagged no-fulltext-access)
• downstream consistency check needed (main agent): hallmarks/cellular-senescence.md, molecules/proteins/p16.md, molecules/proteins/rb.md