R29

log/R29.md — Round 29 entries

Sub-file of log — see parent for index.

[2026-05-07] verify — molecules/proteins/nrf2.md

Pages verified: 1

• molecules/proteins/nrf2.md — corrections: 7 substantive (see below)

Sources checked:

• Itoh 1997 BBRC (10.1006/bbrc.1997.6943) — closed OA (not_oa); tagged no-fulltext-access; DOI+title confirmed via Crossref
• Kobayashi 2004 MCB (10.1128/MCB.24.16.7130-7139.2004) — green OA, download pending; DOI+title confirmed via Crossref; claims corroborated by Yamamoto 2018 PMC review
• Yamamoto 2018 Physiol Rev (10.1152/physrev.00023.2017) — full text verified via PMC9762786 HTML; PMID 29717933
• Shilovsky 2022 (10.1134/S0006297922010060) — closed OA (not_oa); tagged no-fulltext-access
• Narasimhan 2016 (10.3389/fphys.2016.00241) — gold OA, download pending; not fully verified
• Lynch 2019 acn3.660 — abstract verified via PMID 30656180
• Lynch 2021 ana.25934 — full text verified via PMC7894504; PMID 33068037
• Pergola 2011 (10.1056/NEJMoa1105351) — abstract verified via PMID 21699484
• de Zeeuw 2013 BEACON (10.1056/NEJMoa1306033) — full text verified via PMC4496027; PMID 24206459
• UniProt Q16236: length 605 aa, DLG 29-31, ETGE 79-82, bZIP 497-560, Lys596/599 acetylation, Ser40 phosphorylation — all confirmed via REST API
• GenAge ID 283 for NFE2L2 — confirmed via HAGR web
• Open Targets ENSG00000116044 — tier-1 (Approved Drug SM) confirmed via Platform GraphQL API
• KEAP1 27-Cys count — confirmed from UniProt Q14145 sequence
• ClinicalTrials.gov v2 — 6 omaveloxolone active trials confirmed; 0 bardoxolone aging trials

Corrections made:

1. Lynch 2019 design: “Phase 1/2 open-label” → “Phase 2 randomized (3:1 drug:placebo) dose-ranging.” Primary outcome (peak workload) was non-significant. mFARS improvement was secondary outcome, only significant in pes-cavus-free subgroup vs placebo (p=0.01).
2. MOXIe section renamed from “Phase 3” to “registrational Phase 2” per paper’s own designation. Added FAS n (40+42=82) vs randomized n (51+52=103) distinction. Age range corrected to 16–40 for trial (FDA label ≥16).
3. Pergola 2011: “+8.2 mL/min/1.73 m² at 52 weeks” → “+8.2 mL/min/1.73 m² at 24 weeks (primary endpoint).” 52-week results added separately: +5.8 (25 mg), +10.5 (75 mg), +9.3 (150 mg) mL/min/1.73 m².
4. BEACON: Termination reason clarified. Added specific HR for HF hospitalization/death (HR 1.83; 95% CI 1.32–2.55; p<0.001). Clarified “fluid retention” is proposed mechanism in paper, not the termination label.
5. Cancer paradox: KEAP1 NSCLC frequency “~20%” → “~17-20%” with TCGA lung adenocarcinoma data (17.8% in Caucasian cohort; ethnic variation noted).
6. Domain table: Neh7 marked as NOT from Yamamoto 2018 (which covers only 6 domains); attribution to Wang et al. 2013 (PMID 23530057) added. Neh1 corrected to ~435-560 with UniProt bZIP core 497-560 noted. ETGE/DLG positions from UniProt added to Neh2 row.
7. All footnotes updated: corrected design descriptors; replaced “download pending” with PMIDs/PMC IDs; added BEAM trial name for Pergola 2011; added FAS detail to MOXIe footnote; added full quantitative BEACON results.

Unverifiable claims:

• Shilovsky 2022 and Itoh 1997 — closed access; tagged no-fulltext-access in footnotes
• GenAge 283 functional claims (progeria/CR findings) — cited to GenAge database directly, not a specific primary source; not independently verified from primary papers
• KEAP1 ~20 min half-life for Nrf2 — consistent with literature consensus but source not checked against a primary kinetics paper

Downstream propagation needed:

• Any other wiki pages citing the “Phase 3” label for MOXIe (check compound/intervention pages for omaveloxolone)
• Any pages citing “+8.2 mL/min at 52 weeks” for bardoxolone/Pergola 2011

Final verified state: verified: true (partial scope: Itoh 1997 and Shilovsky 2022 full-text unverified; Narasimhan 2016 and Kobayashi 2004 downloads pending)

---

[2026-05-07] verify — molecules/proteins/miro2.md

Pages verified: 1

• molecules/proteins/miro2.md — corrections: 8 substantive (see below)

Sources checked:

• Wang 2011 Cell (10.1016/j.cell.2011.10.018) — local PDF, verified end-to-end
• Fransson 2003 JBC (10.1074/jbc.M208609200) — local PDF, verified end-to-end
• Birsa 2014 JBC (10.1074/jbc.M114.563031) — downloaded this session, verified end-to-end
• López-Domènech 2018 EMBO J (10.15252/embj.201696380) — downloaded this session, verified end-to-end
• Klosowiak 2013 EMBO Rep (10.1038/embor.2013.151) — PDF download stalled; verified from PubMed abstract (PMID 24071720); body claims supported

Corrections made:

1. Sequence identity ~59% → ~60% (Fransson 2003 and López-Domènech 2018 both state “60%”)
2. Birsa 2014 K27/Parkin-Ser65 attribution: corrected major over-attribution — the paper studied Miro1 throughout; Miro2 is mentioned once parenthetically. K27 linkage and Parkin Ser65 dependence are NOT independently demonstrated for MIRO2. Body and footnote updated with explicit attribution note.
3. MIRO2 null mouse phenotype: “Rhot2-null: perinatal lethal or severe” → “viable and fertile to adulthood” (López-Domènech 2018 explicitly states this; Miro-DKO is lethal at E10.5)
4. Fransson 2003 NTPase attribution: removed false attribution of NTPase activity measurements to Fransson 2003, which characterized domain architecture only; tagged unsourced for the C-terminal NTPase claim
5. López-Domènech 2018 protrusion claim: “MIRO2 required for protrusion localization” → corrected to reflect DKO phenotype; Miro2 has more prominent role in actin/Myo19-based distribution but neither single-KO recapitulates DKO
6. phospho-dead MIRO2 claim: clarified that Wang 2011 tested Miro S156A only for Miro1; MIRO2-specific phospho-dead result is extrapolated, not directly shown
7. key-ptms frontmatter: removed unverified MIRO2-specific lysine residues (K96/K119/K164 are from Wang 2011’s Miro studies but not confirmed as MIRO2-specific sites)
8. Added Birsa 2014 finding that Miro1 S156A does not substantially affect ubiquitination kinetics in SH-SY5Y cells — creates uncertainty about whether MIRO2 phosphorylation at the equivalent site is required

Downstream pages potentially affected:

• molecules/proteins/miro1.md — MIRO1 page already verified; the K27-chain attribution in miro1.md (attributed to Birsa 2014) should be correct since Birsa 2014 did study Miro1. Cross-check the K48/K63 chain language — miro1.md says “K48/K63 chains” for MIRO1, but Birsa 2014 shows K27-predominant for Miro1. May need update.
• pathways/pink1-parkin-pathway.md — if it cites MIRO2-specific phosphorylation as independently confirmed
• processes/mitophagy.md — if it cites the K27/Parkin-Ser65/MIRO2 link

[2026-05-07] verify — processes/mitohormesis.md

Pages verified: 1

• processes/mitohormesis.md — corrections: 6 substantive (see below)

Sources checked:

• Ristow 2009 PNAS (10.1073/pnas.0903485106) — local PDF, verified end-to-end
• Houtkooper 2013 Nature (10.1038/nature12188) — local PDF, verified end-to-end
• Houtkooper 2013 Cell (10.1016/j.cell.2013.06.016) — local PDF, verified end-to-end
• Durieux 2011 Cell (10.1016/j.cell.2010.12.016) — downloaded and verified
• Yun & Finkel 2014 Cell Metab (10.1016/j.cmet.2014.01.011) — downloaded and verified
• Lee 2015 Cell Metab MOTS-c (10.1016/j.cmet.2015.02.009) — downloaded and verified
• Ristow & Schmeisser 2014 Dose-Response (10.2203/dose-response.13-035.Ristow) — downloaded and verified
• Tapia 2006 Med Hypotheses (10.1016/j.mehy.2005.09.009) — DOI confirmed via PubMed
• Ristow & Schmeisser 2011 FRBM (10.1016/j.freeradbiomed.2011.05.010) — download failed (hybrid OA, no PMC URL); tagged no-fulltext-access

Corrections made:

1. Houtkooper 2013 Nature organism list: “C. elegans, D. melanogaster, mouse” → “C. elegans, mouse hepatocytes/cell lines” — Drosophila is NOT in this paper’s experimental data
2. Body text: “worms, flies, and mice” → “worms and mice (hepatocytes)” — same Drosophila error
3. UPR^mt ROS-independence: added clarification that Houtkooper 2013 Nature explicitly demonstrates UPR^mt longevity is ROS-independent (NAC did not rescue), separating the two mitohormetic arms
4. DELE1-HRI misattribution: removed [^durieux2011] attribution for DELE1-HRI-ISR mechanism; correct sources are Fessler et al. 2020 and Guo et al. 2020 (Science); tagged for future verification pass
5. Ristow 2009 p-value: “p<0.05” → “p<0.001 by ANOVA” for the primary antioxidant-blockade finding on GIR
6. Houtkooper 2013 Cell SIRT1/SIRT3: “SIRT1/SIRT3” → “SIRT1” only — SIRT3 is not in this paper
7. Tapia 2006 citation: added [^tapia2006] footnote with confirmed DOI (10.1016/j.mehy.2005.09.009) to unsourced in-text attribution
8. All footnote archive statuses updated to reflect actual download state

Pages unverifiable:

• Ristow & Schmeisser 2011 FRBM — tagged no-fulltext-access in footnote

Downstream propagation needed:

• molecules/proteins/pgc-1alpha.md — cites Ristow 2009 in its ROS/mitohormesis section; p-value correction (p<0.001 not p<0.05) should be propagated
• processes/unfolded-protein-response.md — may reference Durieux 2011 for DELE1-HRI; check for same misattribution

[2026-05-07] verify — molecules/proteins/nrf1.md

Pages verified: 1

• molecules/proteins/nrf1.md — corrections: 6 substantive

Source status:

• Wu 1999 Cell (10.1016/s0092-8674(00)80611-x) — local PDF, verified end-to-end
• Huo 2001 MCB (10.1128/MCB.21.2.644-654.2001) — pending → downloaded (PMC86640), verified
• Lee 2024 Aging Cell (10.1111/acel.14446) — pending → downloaded (PMC11984659), verified
• Evans 1990 Genes & Dev (10.1101/gad.4.6.1023) — diamond OA but download failed (CSHL HTTP 520); tagged no-fulltext-access
• Baar 2002 FASEB J (10.1096/fj.02-0367com) — closed-access (not_oa); tagged no-fulltext-access
• Scarpulla 2008 Physiol Rev (10.1152/physrev.00025.2007) — closed-access (not_oa); tagged no-fulltext-access
• UniProt Q16656 — verified via REST API: length 503 aa confirmed; all domain boundaries (Dimerization 1–78, NLS 88–116, DBD 109–305, TAD 301–476) confirmed exact
• Open Targets ENSG00000106459 — druggability tier 4 confirmed (all SM/AB/OC tractability labels false)
• GenAge — NRF1 null entry confirmed; genage-id: null correct

Corrections made:

1. PGC-1α/NRF1 coactivation mechanism — wiki stated PGC-1α “recruits CBP/p300 and SRC-1 to amplify transcription.” Wu 1999 does not describe CBP/p300 or SRC-1 in the NRF-1 coactivation context (those cofactors appear in Puigserver 1998 for PPARγ). Replaced with the actual Wu 1999 mechanism: physical interaction through PGC-1 aa 180–403 and NRF-1 aa 108–143 (within the DNA-binding domain).
2. Wu 1999 footnote organism/model — footnote incorrectly said “n=multiple cell lines + C57BL/6 mice · in-vitro + in-vivo.” Wu 1999 is primarily in-vitro (C2C12, BALB/c 3T3, COS cells); C57BL/6 blastocysts used only for ES clone screening (background from Puigserver 1998), not a direct aging or in-vivo outcome experiment. Footnote corrected to accurately describe assay types and key quantitative results (mtTFA promoter ~4-fold; mtDNA +80%; NRF-1/DN blocks biogenesis to ~10% above control).
3. POLG/Twinkle target-gene attribution — wiki listed these as Function targets with implicit citation to wu1999. Wu 1999 does not mention POLG or Twinkle. Added needs-replication and explicit attribution to Scarpulla 2008 review (closed-access).
4. Huo 2001 mtDNA quantification added — embryonic lethal section now includes the source-stated mtDNA reduction range (30 to <5% of wild-type) and confirms depletion is not caused by apoptosis (TUNEL negative). Huo 2001 footnote updated with strain detail (129/SvJ ES cells / C57BL/6 blastocysts) and precise Mendelian genetics (18/68 blastocysts at 3.5 dpc, zero −/− at 6.5–8.5 or in 412 newborns).
5. Lee 2024 quantitative detail added — body text updated with specific outcomes (OCR/ECAR ratio, ATP, SA-β-gal, p53/p21/p16); GSEA result added (REACTOME_CELLULAR_SENESCENCE NES = −1.555, p = 0.006); footnote updated with full author list, dual senescence model detail, and corrected publication year (2025, published online Dec 2024).
6. Baar 2002 exercise claim flagged — the “~twofold PGC-1α at 18h post-exercise” claim is from a closed-access paper; added no-fulltext-access inline.

Pages unverifiable (closed-access or download failure):

• Evans 1990 (10.1101/gad.4.6.1023) — diamond OA; download failed HTTP 520; claims about NRF-1 founding characterization unverified against full text
• Baar 2002 (10.1096/fj.02-0367com) — closed-access; exercise quantitative claims unverified
• Scarpulla 2008 (10.1152/physrev.00025.2007) — closed-access; POLG/Twinkle target claims unverified

Downstream pages to inspect (main agent):

• molecules/proteins/pgc-1alpha.md — may cite Wu 1999 for CBP/p300 recruitment in NRF-1 context; confirm attribution is limited to PPARγ interaction (Puigserver 1998), not NRF-1
• processes/mitochondrial-biogenesis.md — cites wu1999 for NRF-1 coactivation mechanism; already verified separately; may benefit from adding the interaction domain specifics now available

[2026-05-07] verify — processes/mtdna.md

Pages verified: 1

• processes/mtdna.md — corrections: 5 substantive (see below)

Source status:

• Larsson 1998 (10.1038/ng0398-231) — local PDF, verified
• Lee 2015 (10.1016/j.cmet.2015.02.009) — pending → downloaded, verified
• West 2015 (10.1038/nature14156) — pending → downloaded, verified
• Trifunovic 2004 (10.1038/nature02517) — cross-checked for consistency with oxphos.md; consistent; no re-read needed
• Anderson 1981, Tajima 2002, Wallace 2010 — closed-access (not_oa); tagged no-fulltext-access in footnotes

Corrections made:

1. Larsson 1998 TFAM-KO description — wiki claimed “conditional TFAM inactivation in cardiac muscle causes dilated cardiomyopathy.” That is from Wang 1999 (a subsequent experiment), not Larsson 1998. Larsson 1998 shows: homozygous germline KO → embryonic lethality (E8.5–E10.5) with near-complete mtDNA depletion; heterozygous (+/−) → ~34 ± 7% mtDNA copy number reduction + respiratory-chain deficiency in heart (CI/III/IV/V reduced; CII and citrate synthase normal). No dilated cardiomyopathy is described in Larsson 1998 itself. Body text and footnote corrected accordingly; Wang 1999 noted parenthetically.
2. MOTS-c pathway direction — wiki stated “AMPK → AICAR → ZMP pathway” (reversed). Actual Lee 2015 mechanism: folate cycle inhibition → 5Me-THF depletion → de novo purine synthesis blockade → AICAR accumulation (>20-fold) → ZMP → AMPK activation. Arrow direction and description corrected throughout.
3. MOTS-c ORF details added — GenBank accession KP715230; confirmed 51-bp sORF within MT-RNR1; CD-1 mice (8 weeks, HFD experiment) and C57BL/6 mice (age-decline experiment) used. Dose for in vivo: 5 mg/kg/day IP × 7 days.
4. West 2015 signaling axis — wiki incorrectly attributed downstream signaling to “NF-κB.” West 2015 specifies cGAS → STING → TBK1 → IRF3; ISG mRNAs reduced 70–90% without STING; IRF3 is the effector (not NF-κB). Corrected; aging-inflammation inference flagged as discussion-level (not a direct experimental finding of the paper).
5. Aging-mtDNA gap tag — added needs-human-replication to the West 2015 aging-inflammation claim to accurately reflect that the paper’s primary experimental context is antiviral priming, not normal aging.

Pages unverifiable (closed-access):

• Anderson 1981 (10.1038/290457a0) — genome structure claims (16,569 bp; 37 genes; 13/22/2 breakdown) are textbook-canonical and corroborated by NCBI RefSeq NC_012920; no-fulltext-access tagged in footnote
• Tajima 2002 — humanin neuroprotection claim; no-fulltext-access tagged
• Wallace 2010 — review; copy-number/mutation synthesis; no-fulltext-access tagged

Downstream pages to inspect (main agent):

• molecules/proteins/tfam.md — body describes TFAM KO phenotype; may need to note Wang 1999 vs Larsson 1998 distinction for dilated cardiomyopathy claim
• processes/cgas-sting.md (if it exists) — may inherit the NF-κB vs IRF3 attribution error from mtdna.md seeder
• hallmarks/mitochondrial-dysfunction.md — may cite mtDNA–cGAS-STING axis; check IRF3 vs NF-κB attribution

[2026-05-07] R29 — Telomere biology completion (DONE; 10/10 pages seeded + verified)

Round 29 status: COMPLETE. Triggered by 2026-05-07 lint pass which identified telomere-attrition as the lowest-mechanistic-depth hallmark (4 prot / 0 path / 0 proc) and lowest intervention count (2). All 10 R29 pages now ship verified: true.

Pages seeded + verified (10):

R29a sub-batch (6, parallel seed → parallel verify):

• pathways/telomerase-pathway.md — 14 inbound; central pathway page; KEGG null (no standalone hsa entry), Reactome R-HSA-157579; druggability tier 2 with directional asymmetry (activation tier 3-4, inhibition tier 1 via imetelstat)
• processes/replicative-senescence.md — Hayflick-limit bridge process linking telomere-attrition → cellular-senescence; distinguishes from broader senescence; M1/M2 mortality stages, TIF assay, ALT pathway
• molecules/proteins/terc.md — telomerase RNA component; is-noncoding-rna: true; druggability tier 1 (imetelstat ASO target — schema escalation: ncRNA + ASO drug class needs CLAUDE.md clarification)
• molecules/proteins/trf1.md — shelterin double-strand binding; negative regulator of telomere length; conditional KO causes severe stem-cell exhaustion in stratified epithelia (NOT intestinal — Martínez 2009 uses K5-Cre)
• molecules/proteins/trf2.md — shelterin t-loop / end-protection; ATM-DDR suppression; van Steensel 1998 dominant-negative is HTC75 (not primary fibroblasts)
• molecules/proteins/pot1.md — shelterin single-strand binding; ATR-DDR suppression; mouse Pot1a/Pot1b paralogs; familial CLL/melanoma/glioma POT1 mutations cause elongated telomeres (not shortened)

R29b sub-batch (4, parallel seed → parallel verify):

• molecules/proteins/tin2.md — shelterin scaffold; TINF2 mutations are #1 hotspot for severe DC/Revesz; isoform 1 (TIN2L) = 451 aa canonical (verifier corrected reversed naming)
• molecules/proteins/tpp1.md — telomerase-recruiter; gene ACD; TEL patch is a 7-residue surface (E168/E169/E171/R180/L183/L212/E215) contacting TERT TEN domain; TERT K78 ↔ TPP1 E215 charge-swap pair
• molecules/proteins/dkc1.md — H/ACA RNP catalytic core; X-linked DC; Dkc1 hypomorph mouse (Ruggero 2003) shows pseudouridylation defects BEFORE telomere shortening
• interventions/pharmacological/telomerase-activators.md — class page; covers TA-65, danazol/androgens (FDA approval for DC), TERT mRNA; cancer risk is the central translation barrier; clinical-trials-active: 1 (NCT03312400 RECRUITING danazol)

R29 verification statistics:

• 10 pages verified end-to-end
• ~80 source-text-level corrections applied across the 10 pages
• ~20 scientifically critical corrections, including:
  • POT1 verifier: Pot1b null phenotype direction REVERSED (shortened → elongated G-overhangs; the seeder had it backwards)
  • TPP1 verifier: K170Δ knock-in mouse model claim was FABRICATED (Kocak 2014 is a human genetic study; no mouse). Removed.
  • telomerase-pathway verifier: TCAB1/Cajal-body framing REVERSED (Vogan 2016 shows Cajal localization is NOT required for telomere maintenance; was framed as “rate-limiting checkpoint”)
  • TIN2 verifier: Kim 1999 regulation direction INVERTED (wild-type TIN2 is NEGATIVE regulator; only N-terminal-deletion mutants elongate telomeres)
  • replicative-senescence verifier: Bodnar 1998 delivery method WRONG (was “retroviral”; actually electroporation with MPSV/SV40 plasmids); checkpoint-rescue source reattributed Herbig 2004 → d’Adda di Fagagna 2003
  • TRF1 verifier: Martínez 2009 intestinal crypt claim REMOVED (K5-Cre doesn’t target intestine); PARylation citation moved van Steensel 1997 → Smith 1998
  • TERC verifier: Blasco 1997 multi-tissue phenotype attribution REMOVED (in vivo phenotypes are from Lee 1998 / Rudolph 1999; Blasco 1997 was a cell/tumor study)
  • telomerase-activators verifier: Khincha 2018 DOI was COMPLETELY WRONG (10.1182/bloodadvances.2018018093 was an unrelated blinatumomab case report; correct is 10.1182/bloodadvances.2018016964); Townsley 2016 stopping body NHLBI IRB (not DSMB)
• ~17 wrong DOIs in seeder briefs caught and corrected (TIN2 alone: 4; DKC1: 2; TPP1: 2; TERC: 2; TRF2: 1; POT1: 1; telomerase-activators: 1; consistent with the “seeder-brief DOI memory unreliable” memory item — the wiki-seeder workflow systematically catches these via PubMed efetch + Crossref + DOI lookup cross-check)
• 4 fabricated/unsupported claims caught: K170Δ TPP1 mouse, TCAB1 “rate-limiting”, TRF2 skin-keratinocyte age-decline (left as unsourced — could not source), TRF2+/- lymphoma (left as unsourced — could not source)
• 2 reversed-direction errors: Pot1b overhang direction; Kim 1999 TIN2 regulation direction

Schema escalations surfaced (for next CLAUDE.md cleanup):

• Directionally-asymmetric druggability tier: pathways where activation vs inhibition sit at very different tiers in different clinical contexts (telomerase-pathway: activation tier 3-4 / inhibition tier 1 via imetelstat for MDS). The mtor prototype has a unified tier; pattern may recur (p53-pathway, others).
• ncRNA + ASO drug class druggability: TERC druggability-tier: 1 (imetelstat directly base-pairs with TERC template) is the first ncRNA entity in the wiki with an approved ASO drug. Schema does not address whether oligonucleotide drugs targeting RNAs count as tier 1; needs explicit clarification.
• clinical-stage: enum format inconsistency: telomerase-activators uses phase-2 per CLAUDE.md schema; senolytics uses phase-2-trials. Should be normalized.

Cross-cutting propagation done by main agent (R29 post-verification pass):

• hallmarks/telomere-attrition.md — Bernardes 2012 strain “>95% C57BL/6 background” (was “C57BL/6”); p-values updated (p=0.02, p=0.05); subunit table wikilinked to new R29 protein pages; TPP1 TEL-patch 7-residue list added inline; “Stub pages needed” line updated (only telomere remains unseeded); telomerase paragraph wikilinks TERT/TERC/DKC1/telomerase-pathway/replicative-senescence; telomere-attrition.causes self-loop fixed (was listing itself; removed).
• molecules/proteins/tert.md — complex-subunits: UniProt accessions corrected (NOP10 P57077→Q9NPE3, GAR1 Q9Y285→Q9NY12; were wrong from original seeder pass and survived TERT verification because complex-subunit accessions weren’t re-checked); TPP1 TEL-patch implicit-stub annotation resolved with 7-residue list and Schmidt 2014 K78↔E215 charge-swap mechanism; new Nandakumar 2012 + Schmidt 2014 footnotes added; cross-references and stub-list annotations updated for R29-seeded pages; Bernardes 2012 footnote updated to “>95% C57BL/6 background” + p-value precision; TCAB1 row updated with Vogan 2016 caveat.
• molecules/proteins/terc.md — complex-subunits: UniProt accessions corrected (same NOP10/GAR1 fix as TERT page).
• molecules/proteins/shelterin.md — subunit table wikilinked to all 5 new R29 protein pages; TIN2 row corrected (isoform 1 = TIN2L canonical); TPP1 row updated with TEL-patch residue list + processivity range; TPP1-POT1 mechanism paragraph expanded with full TEL-patch detail + Schmidt 2014 K78↔E215 charge swap + Wang 2007 / Latrick 2010 processivity numbers; new Nandakumar 2012 + Schmidt 2014 + Wang 2007 footnotes added; “individual subunit pages” gap closed.

Hallmark-level impact:

• telomere-attrition mechanistic depth before R29: 4 proteins / 0 pathways / 0 processes / 2 interventions (lint table 2026-05-07).
• telomere-attrition mechanistic depth after R29: ~12 proteins / 1 pathway / 1 process / 3 interventions. Now comparable to epigenetic-alterations (also under-covered; R31 target).

Remaining gaps in this cluster (deferred to future rounds):

• telomere entity page (telomere DNA structure, T-loop biology, measurement methods) — distinct from telomerase machinery; remaining R29 carryover.
• tcab1 / WRAP53 protein page — Cajal-body targeting subunit; flagged on TERT page.
• ta-65 / danazol compound pages — referenced from telomerase-activators class page; class-page has the data but compound pages would absorb per-drug detail.
• idiopathic-pulmonary-fibrosis / dyskeratosis-congenita phenotype pages — referenced from multiple R29 pages.

User direction: R29 was launched 2026-05-07 in response to the user’s request for roadmap rounds focused on improving causes + interventions for the hallmarks. R30 (mitochondrial dynamics) is documented in ROADMAP.md and is the natural next round.

---