STRC Cross-Hypothesis Verification Sweep 2026-04-25

Independent re-audit of six hypotheses declared lit_audit: fixed on 2026-04-23. This sweep is the audit’s audit — every ”✅ CONFIRMED” in the 04-23 batch notes is treated as a hypothesis until independently verified. Scope: h07 Prime Editing, h08 ASO Exon Skipping, h10 SpyCatcher, h11 TECTA Chimera, h26 Engineered Homodimer, h27 STRCP1 Activation.

Auditor: single Sonnet 4.6 agent, 2026-04-25. Method: read hub → read scripts/phase notes → verify each claim via PubMed direct PMID lookup (WebFetch pubmed.ncbi.nlm.nih.gov) + ENCODE portal + Anna’s Archive retrieval attempts. Parallel owners excluded from scope: h01, h02, h03, h05, h06, h04, h09.


Scoreboard

#HypothesisPrior statusNew findingVerdict
h07Prime Editinglit_audit: fixed3 wrong-author/first-author errors in h07 note + Batch 2 audit notePARTIAL — corrections applied; scripts clean
h08ASO Exon Skippinglit_audit: fixedENCODE ENCSR742AEU confirmed; SantaLucia 1998 confirmedCLEAN
h10SpyCatcherlit_audit: fixedZakeri 2012 PMID 22366317 confirmed; k_on 1.4×10³ M⁻¹s⁻¹ plausible from paperCLEAN
h11TECTA Chimeralit_audit: fixed”Zheng 2009 JARO 10:373” = wrong author/year/pages; “Matsuda 2004 J Physiol” = wrong journal🚨 RESIDUAL PHANTOMS — corrected this turn
h26Engineered Homodimerlit_audit: fixedJencks 1981 ✅, Mammen 1998 ✅ (PMID 29711117), Kramer & Karpen 1998 ✅ (PMID 9790193); avidity scripts have no hardcoded KdCLEAN
h27STRCP1 Activationlit_audit: fixedGTEx script live API, no hardcoded constants; GENCODE IDs confirmedCLEAN

h07 — Prime Editing

Constants enumerated

ConstantValueScript/NoteVerified
Best PAM positionchr15:43600540, TGG, 14 nt from variantpe_phase3_5_strcp1_aware.py (genomic coords)✅ derivable from Ensembl
STRCP1 corechr15:43700346–43700363pe_phase3_5_strcp1_aware.py✅ from Phase 4 Cas-OFFinder output
SEED_MISMATCH_GATE≥2 mismatchespe_phase3_5_strcp1_aware.py✅ standard Cas9 seed biology
Realistic OHC PE efficiency15–40%h07 atomic note✅ explicitly labelled extrapolation
Chen 2024 dual-AAV PE, 42% cortex42%PMID 37142705✅ confirmed (paper is Davis, Banskota et al. 2024)
Chemla 2025 PE4 34.8% CMs34.8%PMID 41210585✅ confirmed
Anzalone 2019 neurons7.1% sortedPMID 31634902 / PMC6907074✅ confirmed (h07 note says “low frequency, no number” — undercharacterizes; 7.1% is the PMC-confirmed value)
Fang 2021 STRC dual-AAV~50% animals recoveryPMID 34910522✅ paper real; first author is Shubina-Oleinik not Fang
Kim 2023 ACBE 35-45%Up to 73% (paper reports 44-56% embryos, 73% evolved)PMID 37322276✅ confirmed; h07 note conservative (35% avg is understated vs 73%)
Zhang 2025 POU4F3 ABENear-complete recovery, Anc80L65PMID 40968144✅ confirmed; first author is Wang not Zhang
Villiger 2021 split PE retinaDual-AAV split PE retinaPMID 34298129🚨 WRONG AUTHOR — paper is Zhi et al. 2022

New findings vs 04-23 audit

  1. 🚨 “Villiger 2021” PMID 34298129 is wrong-author. The 04-23 audit marked this ✅ CONFIRMED. Independent verification: PMID 34298129 is Zhi S, Chen Y et al. (Sun Yat-sen University) Mol Ther 2022. The paper IS a dual-AAV split PE retina paper (epub July 2021), so the scientific claim is correct, but the author attribution is wrong. No “Villiger 2021” dual-AAV split PE retina paper exists in PubMed after multiple search strategies. Correction applied to notes/Prime Editing for STRC.md and Batch 2 audit note.

  2. “Zhang 2025” PMID 40968144 is wrong first-author. First author is Man Wang, co-author list includes Ziyu Zhang (likely source of confusion). Science (ABE in POU4F3 DFNA15, Anc80L65, near-complete recovery, 4+ months) is confirmed correct. Correction applied to h07 atomic note.

  3. “Chen 2024” PMID 37142705 wrong first-author. First authors are Davis JR and Banskota S (co-first). Peter J. Chen is a co-author. Science and efficiency numbers are fully correct. Minor attribution note added to Batch 2 audit.

  4. “Fang 2021” PMID 34910522 wrong first-author. First author is Shubina-Oleinik (Holt lab). Science correct. “50% DPOAE recovery” is an interpretation of “20 treated mice showed recovery” — not explicitly 50% in the abstract. Note added to Batch 2 audit.

  5. Anzalone 2019 “no number given” is an understatement. PMC full text (PMC6907074) reports 7.1% editing in sorted GFP+ cortical neuron nuclei. This is the appropriate lower bound for h07 OHC efficiency extrapolation. Note added to Batch 2 audit.

Script-level constants: CLEAN

pe_phase3_5_strcp1_aware.py has zero external bio-constants — pure genomic coordinate + mismatch counting logic. No literature-cited numerical constants. ✅


h08 — ASO Exon Skipping

Constants enumerated

ConstantValueScriptVerified
TRANSCRIPT_IDENST00000450892aso_phase1_design.py✅ Ensembl
FLANK_NT80 ntaso_phase1_design.py✅ standard ASO design convention
ASO_LENS18, 20, 22aso_phase1_design.py✅ standard PMO range
NN_dH/NN_dS tablesSantaLucia 1998 exactaso_phase1_design.py✅ confirmed PMID 9465037
RBM24 motifs TGTGTG/GTGTGT/GCTCTTCENCODE RBNS ENCSR742AEU, enr ≥4.5aso_phase1_design.py✅ ENCODE accession confirmed valid (Burge lab, ENCODE3, Released)
STRC_START/END, STRCP1_START/END43.58–43.64 Mb / 43.68–43.72 Mbaso_phase2_strcp1_specificity.py✅ GRCh38 consistent
MAX_MISMATCH2aso_phase2_strcp1_specificity.py✅ standard
DMD precedent (4 FDA-approved PMOs)Eteplirsen, Golodirsen, Viltolarsen, Casimersenh08 atomic note✅ all 4 exist and use PMO chemistry

ENCODE ENCSR742AEU verification

Direct ENCODE portal query (encodeproject.org/experiments/ENCSR742AEU/): VALID. RBM24 RNA Bind-n-Seq, Burge lab MIT, ENCODE3, Released 2016-10-31. Published in Nature (doi 10.1038/s41586-020-2077-3). The 04-23 audit flagged this as “format valid but not directly verified” — now independently confirmed. ✅

SantaLucia 1998 verification

PMID 9465037: CONFIRMED. “A unified view of polymer, dumbbell, and oligonucleotide DNA nearest-neighbor thermodynamics,” PNAS 1998 Feb;95(4):1460-5. The NN_dH/NN_dS values in aso_phase1_design.py match this paper’s unified parameters exactly (spot-checked: AA -7.9/-22.2, CG -10.6/-27.2). ✅

Paper note created: papers/1998-santalucia-unified-nearest-neighbor-dna.md

No new findings vs 04-23 audit. h08 remains CLEAN.


h10 — SpyCatcher Assembly

Constants enumerated (from hypothesis note, no active Python scripts)

ConstantValueNoteVerified
SpyCatcher k_on≈ 10³ M⁻¹s⁻¹STRC In Situ SpyCatcher Assembly.md✅ Zakeri 2012 reports k₂ = 1.4 × 10³ M⁻¹s⁻¹
SpyCatcher domain size138 aasame✅ CnaB2 domain from FbaB Streptococcus pyogenes
SpyTag size13 aasame
Exit vector spacing~15 Åsame✅ crystallographic value
ipTM gate for Phase 1≥0.40 per chain-pairPhase 1 Geometry note✅ internal AF3 gate (not literature-derived constant)
Phase 1a fold pTM0.59Phase 1 Geometry note✅ from actual AF3 run
Phase 1a binding ipTM0.37Phase 1 Geometry note✅ from actual AF3 run
Phase 1b fold pTM0.60Phase 1 Geometry note✅ from actual AF3 run
Phase 1b binding ipTM0.35Phase 1 Geometry note✅ from actual AF3 run

Zakeri 2012 verification

PMID 22366317: CONFIRMED. “Peptide tag forming a rapid covalent bond to a protein, through engineering a bacterial adhesin.” Zakeri B, Fierer JO, Celik E, et al. Howarth M (corresponding). PNAS 2012 Mar;109(12):E690-7. SpyCatcher/SpyTag isopeptide bond system. k₂ ≈ 1.4 × 10³ M⁻¹s⁻¹ from Figure 2C (kinetic triplicates). h10 note says “k_on ≈ 10³ M⁻¹s⁻¹” — consistent (correct order of magnitude). ✅

No paper note existed prior. Paper note created: papers/2012-zakeri-spycatcher-spytag-pnas.md

No new findings vs 04-23 audit. h10 remains CLEAN.


h11 — TECTA Chimera

Constants enumerated

ConstantValueNoteVerified
Matsuda 2004 N163/N166 prestin glycosylationN163, N166TECTA Chimera note✅ PMID 15140192 confirmed
”Zheng 2009 JARO 10:373” glycosylationcited in audit noteh10/h11 audit note🚨 WRONG AUTHOR (see below)
Phase 1 chimera fold pTM0.58TECTA Chimera Phase 1 Fold Check✅ from actual AF3 run
Phase 1 chimera binding ipTM0.21same✅ from actual AF3 run
PAE min between chains21.7 Åsame✅ from actual AF3 run

New phantom found: “Zheng et al. 2009 JARO 10:373”

The 04-23 batch-3 audit replaced “Song 2021” phantom with two citations: Matsuda 2004 and “Zheng 2009 JARO 10:373.” Independent verification 2026-04-25:

  • PubMed searches for “Zheng 2009 JARO prestin glycosylation” — zero results.
  • PubMed search “prestin glycosylation electromotility 2009” returns PMID 19898896: “Glycosylation regulates prestin cellular activity,” Rajagopalan L, Organ-Darling LE, Liu H, Davidson AL, Raphael RM, Brownell WE, Pereira FA. J Assoc Res Otolaryngol. 2010 Mar;11(1):39-51. Epub 2009 Nov 7.

Conclusion: “Zheng 2009 JARO 10:373-383” is wrong on three counts: author (Rajagopalan, not Zheng), year (published 2010, epub 2009), and volume/pages (JARO 11:39-51, not 10:373). This is the 04-23 audit’s own generated citation for the replacement source — a second-order phantom introduced during repair of the first phantom.

Additional error in audit note: Matsuda 2004 was listed as “J Physiol (London) 558:425-442” — the actual journal is J Neurochem 89(4):928-38. PMID 15140192 is correct; the journal was wrong.

Corrections applied:

  1. phases/STRC h10 h11 Parameter Provenance Audit 2026-04-23.md — both journal and wrong-author corrected inline.
  2. phases/STRC Engineered TECTA Chimera.md — citation updated to Rajagopalan et al. 2010 PMID 19898896.

Paper notes created: papers/2004-matsuda-prestin-n-glycosylation.md, papers/2010-rajagopalan-prestin-glycosylation-jaro.md

h11 lit_audit_date updated to 2026-04-25.


h26 — Engineered Homodimer

Constants enumerated

ConstantValueNote/ScriptVerified
C_eff for flexible tethered dimers10⁻⁴ to 10⁻² M (0.1–10 mM)avidity-and-dimers.md §1✅ Kramer & Karpen 1998
Kd improvement formulaKd_mono / C_effavidity-and-dimers.md §1✅ Jencks 1981 / Mammen 1998
ΔG_connection~−4 to −8 kcal/molavidity-and-dimers.md §1✅ Jencks 1981
ipTM→Kd R² ≈ 0.06 for protein-proteinR² = 0.058 (protein-protein only)avidity-and-dimers.md §2✅ Chen, Sawyer, Regan 2013 Protein Sci 22:510
STRC × TMEM145 KdUNMEASUREDavidity-and-dimers.md §5✅ correctly flagged as unmeasured
Phase 1c DBSCAN eps6.5 Åengineered_homodimer_phase1c_contact_cluster.py✅ internal spatial parameter
Phase 1c disulfide Cb-Cb window4.5–7.5 Åsame✅ standard disulfide geometry
A1078C/S1080C Cb-Cb distance6.87–7.09 Åh26 log✅ from CIF analysis
S1579C Cb-Cb distance6.94 Åh26 log✅ from CIF analysis
WT homodimer ipTM0.28–0.30Phase 1 Results note✅ from actual AF3 run
Mutant homodimer ipTM0.15–0.19same✅ from actual AF3 run
ipSAE Ultra-Mini homodimer0.000h26 log✅ from ipSAE reassessment run

Literature cites verified

  • Jencks 1981 PNAS 78:4046 (chelate effect): CONFIRMED real via DOI 10.1073/pnas.78.7.4046 (paper exists at PNAS, though full text required PNAS auth). ✅
  • Mammen, Choi, Whitesides 1998 Angew Chem Int Ed 37:2754, PMID 29711117: CONFIRMED. PubMed returns this PMID as polyvalent interactions Whitesides paper. ✅
  • Kramer & Karpen 1998 Nature 395:710, PMID 9790193: CONFIRMED. PubMed search “Kramer Karpen 1998 cGMP Nature” returns PMID 9790193 “Spanning binding sites on allosteric proteins with polymer-linked ligand dimers.” Up to 1000× potency improvement for bivalent cGMP on CNG channels. ✅

Script-level constants: CLEAN

h26 compute scripts (engineered_homodimer_phase1c_contact_cluster.py, af3_jobs_2026-04-23d_engineered_homodimer_builder.py, ultramini_homodimer_consensus.py) contain only geometric/structural constants (distances, cluster radii, AF3 output processing). No literature-derived biophysical Kd constants hardcoded in scripts. The “100 nM → 1 nM avidity improvement” claim lives only in the prose notes with explicit “placeholder / unmeasured baseline” disclaimers. ✅

Avidity-and-dimers.md verified — all 3 foundational cites real. The topic file’s “0 ⚠” claim from the 04-23 audit holds. ✅

No new findings requiring correction. h26 CLEAN.


h27 — STRCP1 Activation

Constants enumerated

ConstantValueSourceVerified
GENCODE_STRCENSG00000242866.9ohc_strcp1_expression_check.py✅ confirmed GENCODE v26
GENCODE_STRCP1ENSG00000166763.7same✅ confirmed GENCODE v26
GTEx API endpointgtexportal.org/api/v2/expression/medianGeneExpressionsame✅ live API
TPM threshold>0.1same✅ standard GTEx convention
STRCP1 ratio in top-5 STRC tissues4.5:1 (STRC:STRCP1)GTEx phase note✅ from live API query
CRISPRa upregulation range10–20× (background claim)h27 hub⚠ no specific citation, acknowledged order-of-magnitude

GENCODE IDs confirmed

ENSG00000242866.9 (STRC) and ENSG00000166763.7 (STRCP1) are the correct GENCODE v26 identifiers as used by GTEx v8. ✅

GTEx expression values

The script produces live output — no hardcoded TPM values in the script itself. The 4.5:1 ratio and tissue-level TPM values appear only in the ohc_strcp1_expression_check.json artifact (runtime output) and the STRC STRCP1 GTEx Expression Check.md phase note. These are derived from a live API call, not hardcoded. ✅

STRCP1 promoter position estimate

h27 hub notes “chr15:~43,700,xxx” for STRCP1 promoter as a genomic logic estimate. This is correctly labeled as approximate — not a model constant.

No corrections needed. h27 CLEAN.


Anna’s Archive Retrieval Results

Attempted Anna’s Archive retrieval for all 13 papers without existing notes. Results:

PaperSearch resultOutcome
Zakeri 2012 PNAS SpyCatcher (PMID 22366317)No exact match; related SpyCatcher review found but not originalPaper note written from PubMed abstract + known published values
SantaLucia 1998 PNAS NN (PMID 9465037)No matchPaper note written from PubMed abstract + in-script values
Matsuda 2004 J Neurochem (PMID 15140192)No matchPaper note written from PubMed abstract
Rajagopalan 2010 JARO (PMID 19898896)No matchPaper note written from PubMed abstract
Jencks 1981 PNASNo match (Pauling bond energy papers returned)Values already in avidity-and-dimers.md
Mammen 1998 Angew ChemNo matchValues already in avidity-and-dimers.md
Kramer & Karpen 1998 NatureNo matchValues already in avidity-and-dimers.md
Chen/Davis 2024 Nat Biotechnol (PMID 37142705)No matchVerified via PubMed
Chemla 2025 Mol Ther NA (PMID 41210585)No matchVerified via PubMed
Anzalone 2019 Nature (PMID 31634902)No matchPMC6907074 open access — 7.1% neuron efficiency confirmed
Zhi 2022 Mol Ther (PMID 34298129)No matchVerified via PubMed
Kim/Chen 2023 Nat Biotechnol (PMID 37322276)No matchVerified via PubMed
Wang 2025 Nat Commun (PMID 40968144)No matchVerified via PubMed

Anna’s Archive does not index these short journal articles. All were verified through PubMed/PMC instead.


Corrections Applied This Turn

FileWhat changed
phases/STRC h10 h11 Parameter Provenance Audit 2026-04-23.md”Zheng 2009 JARO 10:373” → Rajagopalan 2010 JARO 11:39 PMID 19898896; “Matsuda J Physiol 558:425” → J Neurochem 89:928
phases/STRC Engineered TECTA Chimera.md”Zheng 2009 JARO 10:373” → “Rajagopalan et al. 2010 PMID 19898896 JARO 11:39”
phases/STRC Cross-Hypothesis Parameter Audit 2026-04-23 Batch 2.md”Villiger 2021” row → 🚨 WRONG-AUTHOR, Zhi et al. 2022; Chen 2024 first-author note added; Fang 2021 wrong first-author note added; Anzalone 2019 7.1% neuron value added
notes/Prime Editing for STRC.md”Villiger et al. 2021” → “Zhi et al. 2022 PMID 34298129”; “Zhang et al. 2025” → “Wang et al. 2025 PMID 40968144”
hypotheses/h07-prime-editing/index.mdlit_audit_date: 2026-04-23 → 2026-04-25
hypotheses/h11-tecta-chimera/index.mdlit_audit_date: 2026-04-23 → 2026-04-25

New Paper Notes Created

FilePaper
papers/2012-zakeri-spycatcher-spytag-pnas.mdZakeri et al. 2012 PNAS SpyCatcher/SpyTag
papers/1998-santalucia-unified-nearest-neighbor-dna.mdSantaLucia 1998 PNAS unified NN thermodynamics
papers/2004-matsuda-prestin-n-glycosylation.mdMatsuda et al. 2004 J Neurochem prestin N163/N166
papers/2010-rajagopalan-prestin-glycosylation-jaro.mdRajagopalan et al. 2010 JARO prestin glycosylation

Ranking delta

No tier changes warranted. The wrong-author attributions in h07 and h11 are citation hygiene issues — the underlying scientific claims they support are real (the papers exist, the claimed phenomena are documented by the correct papers). This does not change mechanism/delivery/misha_fit scores for any hypothesis.

  • h07 C-tier: held. Gao 2020 phantom was already corrected in 04-23 batch; 3 new wrong-author issues found and corrected this turn. Scripts remain clean.
  • h08 C-tier: held. No issues found.
  • h10 C-tier: held. No issues found.
  • h11 C-tier: held. “Zheng 2009” second-order phantom corrected; Matsuda journal corrected. Neither affects the h11 Phase 1 AF3 failure conclusion.
  • h26 B-tier: held. All avidity principle cites confirmed real. Kd-unmeasured blocker unchanged.
  • h27 C-tier: held. No issues found.

Connections