R14

log/R14.md — Round 14 entries

Sub-file of log — see parent for index.

[2026-05-05] Round 14 — Schema additions for synthesis MOCs (DONE)

R14 was the explicit PREREQUISITE for R15 (causality graph) and R16 (intervention matrix). Pure frontmatter-population pass; no body content edits.

CLAUDE.md schema additions

• type: hallmark: added mechanistic-tier: (proximal | intermediate | integrative — finer than López-Otín’s primary/antagonistic/integrative; reflects this wiki’s own intervention-evidence ranking) + intervention-tractability: (high | moderate | low | none) + caused-by: [] + causes: [] (R15 placeholders).
• type: pathway: added druggability-tier: + caused-by: [] + causes: [].
• type: protein: added ensembl:, is-noncoding-rna: false, druggability-tier:, caused-by: [], causes: [].
• type: compound: added biologic: false, who-inn:, translation-gap:, next-experiment:. Documented biologic-compound convention (PubChem CID null acceptable; ChEMBL still tracks).
• type: process: added druggability-tier:, caused-by: [], causes: [].
• type: intervention: added translation-gap:, next-experiment:.
• type: framework: documented schema convention — framework pages do NOT carry verified fields (apoptosenes precedent established R13). Examples: hallmarks-of-aging, sens-damage-categories, apoptosenes.
• type: microbe (NEW): formalized for akkermansia-muciniphila (R12 escalation now resolved). Required fields: ncbi-taxonomy, phylum, family, gram-stain, oxygen-tolerance, host, hallmarks, verified.
• Citation discipline: documented retracted-paper convention (bold “RETRACTED” prefix + retraction DOI in footnote; replacement paper carries the active citation). Precedent: Wenz 2009 → Leick 2010 on pgc-1alpha (R13).

Frontmatter-population batches (6 parallel agents)

Total: 128 pages touched, frontmatter-only (no body edits).

Batch 1 (proteins) — 4 sub-batches in parallel against Open Targets Platform GraphQL API:

Sub-batch	n	Tier 1	Tier 2	Tier 3	Tier 4	Null+gap
A–C (atg/bcl/caspase cluster)	33	10	16	0	6	1 (age-1 C. elegans)
D–I (foxo/igf/insr cluster)	24	7	5	7	3	2 (daf-16, daf-2 C. elegans)
K–P (mdm2/oct4/pi3k/parkin)	25	6	4	9	6	0
R–X (raptor/sirt1/tau/tet2)	16	3	10	1	2	0
Total	98	26 (27%)	35 (36%)	17 (17%)	17 (17%)	3 (3%)

The wiki’s protein coverage is heavily skewed toward clinically-relevant targets (~63% tier 1+2), confirming the aging-intervention focus.

Ensembl ID corrections caught at batch time (3):

• LC3 (MAP1LC3B): brief had ENSG00000197548 (which is ATG7) → corrected to ENSG00000140941 (verified via UniProt Q9GZQ8).
• TBK1: existing ensembl: ENSG00000198793 was RAB17 → corrected to ENSG00000183735 per UniProt Q9UHD2.
• atg101 + bcl2l13: had null ensembl values → replaced with proper IDs (ENSG00000123395, ENSG00000099968).

45 ensembl IDs added across protein pages where the field was previously missing.

Batch 2 (hallmarks) — 12 pages:

Cross-tabulation (mechanistic-tier × intervention-tractability):

	high	moderate	low
proximal	0	1 (epigenetic-alterations)	2 (genomic-instability, telomere-attrition)
intermediate	3 (cellular-senescence, deregulated-nutrient-sensing, disabled-macroautophagy)	2 (loss-of-proteostasis, mitochondrial-dysfunction)	0
integrative	1 (chronic-inflammation)	2 (altered-intercellular-communication, dysbiosis)	1 (stem-cell-exhaustion)

Key finding for R15: “intermediate × high” is the single richest cluster — cellular-senescence + deregulated-nutrient-sensing + disabled-macroautophagy are the three hallmarks where verified clinical interventions exist (D+Q senolytics, rapamycin/CR, spermidine/CR/rapamycin via autophagy). The proximal-damage classes (genomic-instability, telomere-attrition) have low tractability — direct DDR/telomerase-restoration translation has stalled. Chronic-inflammation is the standout integrative-high cluster (CANTOS Phase 3 verified).

Batch 3 (compounds + interventions) — 18 pages:

translation-gap distribution:

Category	Count	Pages
phase-3-rct-needed	6	dasatinib, quercetin, metformin, rapamycin, creatine, senolytics-class
biomarker-only	3	fisetin, NR, spermidine
translation-blocked-safety	2	navitoclax, A-1331852 (BCL-xL platelet toxicity)
human-evidence-strong	2	caloric-restriction, senomorphics-class
effect-size-too-small	2	NMN, urolithin-A
translation-blocked-cost	1	canakinumab
translation-blocked-bioavailability	1	EGCG
mechanism-disputed	1	taurine (Singh 2023 vs Fernandez 2025)
preclinical-only	0	—

Every page has at least Phase 1 human PK or observational cohort data; “preclinical-only” gap empty in current compound coverage. The phase-3-rct-needed plurality (33%) names the wiki’s translation thesis: many compounds with positive Phase 1-2 / mouse data lack hard-endpoint Phase 3 RCTs in aging populations.

lint-pass.md extension

Added an “R14 schema coverage” section that scripts each type-block’s required fields. Future seeders/verifiers will catch missing druggability-tier / mechanistic-tier / intervention-tractability / translation-gap / next-experiment / caused-by / causes fields automatically. Also added a placeholder “Intervention matrix coverage” section for R16’s Dataview self-watchdog blocks.

Schema escalations resolved

• Framework verified field convention: documented (strip from framework pages).
• type: microbe: formalized.
• Biologic compound flag: added (biologic: true + who-inn:).
• Retracted-paper citation convention: documented.

All 4 R12+R13 schema escalations are now closed. R14 success criteria met:

• 98/98 protein pages have druggability-tier ✓
• 12/12 hallmarks have mechanistic-tier + intervention-tractability ✓
• 18/18 compound + intervention pages have translation-gap + next-experiment ✓
• CLAUDE.md updated ✓
• lint-pass.md extended ✓

Outstanding (deferred to R15)

• caused-by: and causes: fields are seeded as [] placeholders on all 12 hallmark + 98 protein + 7 process + 13 pathway pages. Population is R15 work (synthesis-decision, not frontmatter-scrape).
• Spot-check verification of druggability-tier assignments against Open Targets web UI: deferred — agents reported per-page rationale + Open Targets API is deterministic.