AUDIT — h14-protein-replacement (STRC Protein Replacement Therapy)
- Tier: C · mech 2 / deliv 1 / misha_fit 3 ·
next_step: No delivery route - lit_audit before:
deferred(2026-04-23) - lit_audit after:
partial, 2 blockers - Grounding:
GROUNDING/hypc-h14-protein-replacement.md - Proportionality: C-tier, no scripts. Short pass — but it surfaced a real contradiction, so the writeup is longer than the retrieval.
Parameter-provenance table
| # | Claim | Value | Source | Local sources/lit/ | Status |
|---|---|---|---|---|---|
| 1 | STRC full-length mass | ~220 kDa | derived from 1775 aa + heavy N-glycosylation | — | ✅ derived, internally coherent |
| 2 | Mini-STRC mass (700–1775) | ~115 kDa | derived, 1076 aa | — | ✅ derived |
| 3 | Diffusion coefficient at 115 kDa | D ≈ 2–4 µm²/s | ”Stokes-Einstein” — method named, no citation, no radius given | — | ❌ UNSOURCED ESTIMATE |
| 4 | Perilymph length | ~5 mm | anatomical | — | 🔶 uncited but textbook; cf. Salt 2001’s human ST ~28.5 mm |
| 5 | Distribution time to all OHC rows | ”hours” | computed from #3 + #4 | — | 🔶 inherits #3 |
| 6 | ERT scale feasibility | qualitative | ”modern CHO/HEK systems” | — | ⚙ qualitative |
| 7 | N-glycosylation requirement | qualitative | [[STRC N-Glycosylation Analysis]] (in-vault) | — | ⚙ |
| 8 | RWM permeability — the step the kill actually turns on | never computed | Salt & Ma 2001 is already ✅ in sources/lit/ and was never applied here | 2001-salt-ma-quantification-rwm-permeability.md | ❌ NOT APPLIED |
Finding 1 — the hub’s kill contradicts the note’s own conclusion
Hub: next_step: No delivery route.
Note, line 64:
“Diffusion modeling — Stokes-Einstein: at 115 kDa, D ≈ 2-4 µm²/s in aqueous. Perilymph ~5 mm long. Time to diffuse to all OHC rows: hours. Feasible.”
The note concludes distribution is feasible in hours; the hub kills the
hypothesis for having no delivery route. These are almost certainly about
different steps — crossing the round-window membrane into perilymph versus
diffusing along perilymph once inside — and the reconciliation is probably
straightforward. But it is written down nowhere. As the vault stands, a reader
cannot tell whether h14 was killed on evidence or on an unstated assumption, and
lit_audit: deferred meant nobody had ever asked.
For a C/D-tier hypothesis the audit’s whole job is confirming the kill rests on sourced evidence. Here it rests on an unrecorded inference.
Finding 2 — the kill is quantifiable from a paper the vault already owns
sources/lit/2001-salt-ma-quantification-rwm-permeability.md (type: lit) was
retrieved, MinerU-parsed, and closed to ✅ during the h09 audit. Line 44:
| RWM permeability (TMPA) | 1.9 × 10⁻⁸ | cm/s | Guinea pig; TMPA ion MW ~166 Da |
with the paper’s own endorsed MW-scaling method worked through at lines 61–62. This is precisely the RWM-crossing step that h14’s kill turns on, and it has never been applied to h14.
Extending the same MW^−1/3 scaling to mini-STRC at 115 kDa would give P ≈ 2.2 × 10⁻⁹ cm/s, ~10× below the TMPA ion — i.e. it would likely support the kill. I record that as an illustration of the available calculation, not as a result: Salt & Ma measured a 166 Da ion, and pushing Stokes–Einstein across nearly three decades of molecular weight onto a heavily glycosylated 115 kDa glycoprotein is far outside what that paper supports. Hardening that number into a “value” is exactly what §0c exists to prevent, and I am not doing it.
The finding is structural, not numerical: h14’s kill can be grounded in
literature the vault already holds, and nobody has done it. That is a cheap
concrete task, not a literature gap — which is why h14 lands on partial rather
than not-modelable.
Cross-hypothesis note: this makes Salt & Ma 2001 load-bearing for h09 (ototopical ear drop) and h14 (intratympanic protein). One paper, two hypotheses, one already using it properly.
Verdict
lit_audit: partial · lit_audit_date: 2026-08-21 · 2 blockers
lit_audit: partial
lit_audit_date: 2026-08-21
lit_audit_blockers:
- "Kill rationale is unrecorded and contradicts the hypothesis note. Hub says next_step 'No delivery route'; STRC Protein Replacement Therapy.md:64 concludes intracochlear distribution is 'Feasible' in hours, and line 50 argues mini-STRC at 115 kDa is smaller than several approved ERT proteins. The two are probably about different steps (crossing the round-window membrane vs diffusing along perilymph once inside), but that reconciliation is written down nowhere. A reader cannot currently tell whether h14 was killed on evidence or on an unstated assumption. Next step: state the kill's actual step and basis in the hub."
- "Diffusion coefficient D = 2-4 um^2/s at 115 kDa (STRC Protein Replacement Therapy.md:64) — UNSOURCED ESTIMATE. Method is named (Stokes-Einstein) but no citation and no hydrodynamic radius is given, and it is load-bearing for the 'Feasible' half of the contradiction above. Related and more useful: the RWM-crossing step the kill actually turns on is quantifiable from a paper the vault ALREADY OWNS and has never applied here — sources/lit/2001-salt-ma-quantification-rwm-permeability.md (type: lit), RWM permeability 1.9e-8 cm/s for the 166 Da TMPA ion, with the paper's own endorsed MW-scaling method at lines 61-62. Applying it to 115 kDa would likely SUPPORT the kill, but the extrapolation spans ~3 decades of molecular weight onto a heavily glycosylated protein and must be flagged as an extrapolation, not a measurement."Ranking delta
C-tier HELD. No mech/deliv/misha_fit change. See RANKING-DELTA.md.