STRC Dual-Vector vs Single-Vector Transduction
⚠️ Model Update (2026-03-19, recalibrated 2026-04-16)
The numbers in this note reflect v1 Simple Poisson model and are now superseded.
v2 Gamma-Poisson (negative binomial, accounts for cell-to-cell heterogeneity in AAV uptake): see STRC Gamma-Poisson Transduction Model.
- Single-vector: 77.8% OHC transduction (was 67.4%)
- Dual-vector (R=50%): 27.3% (was 1.2%)
- Advantage ratio: 2.8-4.7x (was 56.5x)
v3 OTOF clinical recalibration (2026-04-16) — model recalibrated against OTOF/DB-OTO trial data (2026-04-16-lesperance-otoferlin-gene-therapy; Lustig 2024 Regeneron; Sun et al. 2024 Lancet). Code: ~/STRC/models/dual_vector_otof_calibration.py.
The model underpredicts dual-vector transduction at high titers (mean error 14%) because recombination efficiency is higher in vivo than the bench R=0.50 assumption.
Revised advantage at clinical titer: 2.2x (single 89.8% vs dual 40.4%)
2.2x is still a clinically meaningful advantage — dual-vector requires 2.2x more vector copies per OHC to match single-vector coverage. This is an honest correction and should replace the 2.8-4.7x figure in all communications.
The direction remains the same: single-vector is decisively better. The magnitude is now grounded in clinical data, not bench assumptions.
The Core Question
STRC gene (5,325 bp) exceeds AAV packaging limit (4,700 bp). Current approach (Iranfar 2026): dual-vector — split gene into two halves, two separate AAVs, inject both, hope they recombine in the same cell.
Mini-STRC hypothesis: 3,546 bp coding sequence fits in single AAV. Question: how much better is single-vector?
Answer from the Poisson model: 56.5 times better.
The Math
Viral particle uptake follows a Poisson distribution. If the average number of particles reaching a cell (MOI, multiplicity of infection) is λ, then the probability of getting at least one particle is:
P(at least 1) = 1 - e^(-λ)
For dual-vector: both AAV-A and AAV-B must enter the same cell. If they share the same receptors (AAVR/HSPG), they compete — each gets half the effective MOI. Co-transduction probability:
P(co-transduction) = P(A≥1) × P(B≥1)
Even after co-transduction: the two halves must recombine intracellularly. Published efficiency: 10-30% (overlap or split-intein designs). This loss multiplies on top of the co-transduction penalty.
Calibration from Experimental Data
Calibrated from Omichi et al. (2020):
- Single AAV2 via RWM: 83.9% OHC transduction in mice at 3.75×10¹² GC/mL
- This implies effective delivery rate = 0.12% of injected particles actually reach OHC
- Most particles stay in perilymph, clear via aqueous humor, or are degraded
Source: Omichi et al. (2020). Cochlear gene therapy with dual-AAV. PMC7270144
Human Scaling
Mice → humans: the perilymph volume difference is the critical scaling factor.
| Parameter | Mouse | Human | Scaling |
|---|---|---|---|
| Perilymph volume | 2.5 µL | 191 µL | 76x |
| OHC count | ~2,000 | ~12,000 | 6x |
| Required titer to maintain MOI | baseline | ~13x higher | volume-driven |
At standard clinical titer (3.75×10¹² GC/mL), human scaling reduces effective MOI significantly. The model accounts for this when computing human predictions.
Key Results (Human Prediction at Standard Titer)
Python model: ~/DeepResearch/strc/dual_vs_single_vector.py
Results: ~/DeepResearch/strc/dual_vs_single_results.json
| Metric | Single-vector (mini-STRC) | Dual-vector (full STRC) |
|---|---|---|
| Effective MOI | 1.1 per cell | 0.3 per vector per cell |
| Transduction probability | 67.4% | 1.2% |
| OHC with functional protein | 8,082 of 12,000 | 143 of 12,000 |
| Advantage ratio | — | 56.5x |
At standard clinical titer: single-vector mini-STRC delivers functional protein to 56.5 times more hair cells than dual-vector full STRC.
Dual-vector is further penalized by 10-30% intracellular recombination efficiency, which is already included in the 1.2% figure. Without the recombination penalty, dual-vector would be ~5-10% — still 7-14x worse than single.
Titer Dependence (Sweep)
| Viral titer (GC/mL) | Single-vector | Dual-vector | Gap |
|---|---|---|---|
| 1×10¹¹ | 11.0% | 0.04% | 275x |
| 1×10¹² | 50.3% | 0.5% | 101x |
| 3.75×10¹² | 67.4% | 1.2% | 56.5x |
| 1×10¹³ | 82.1% | 4.9% | 17x |
| 5×10¹³ | 95.0% | 22.4% | 4x |
Key pattern: the gap widens at clinically realistic titers. Only at extreme titers (>10¹³ GC/mL, difficult to manufacture and potentially toxic) does dual approach single efficiency. The sweet spot for real-world therapy (10¹²-10¹³) is where single-vector has the greatest advantage.
The Recombination Problem
Even if both vectors enter the same cell, the two half-genes must find each other and recombine. Published recombination efficiency:
- Overlap design (Ghosh et al. 2011, DMD studies): 10-30% of co-transduced cells achieve functional protein
- Split-intein design (better but more complex): up to 80% in optimized conditions, but AAV-intein constructs are harder to produce
For the baseline (overlap design, 15% recombination):
- P(co-transduction): ~8% at 3.75×10¹² GC/mL (both vectors in same cell)
- P(recombination | co-transduced): 15%
- Net: 8% × 15% = 1.2% of OHCs get functional protein
Receptor Competition Caveat
Model assumes independent Poisson uptake for each vector, which is conservative (optimistic for dual-vector). In reality:
- AAV-A and AAV-B share the same surface receptors (AAVR, HSPG)
- They compete for binding sites at the cell surface
- Steric exclusion: after one capsid binds, it physically blocks the adjacent site
- Experimental data supports this: Omichi 2020 measured 83.9% single vs 65.6% dual (in optimized overlap design), consistent with competition
The real dual-vector efficiency is likely worse than the model predicts. 56.5x advantage is a lower bound.
Connection to Mini-STRC
This analysis assumes mini-STRC is functionally equivalent to full STRC. That assumption needs experimental validation. AF3 Job 5 shows better folding (pTM 0.81 vs 0.63). If mini-STRC has equivalent or better function, the 56.5x transduction advantage translates directly to therapeutic outcome.
If mini-STRC has, say, 80% of full STRC function, the effective advantage is still: 56.5x × 0.80 = 45x — still decisive.
Context: Why This Matters for Misha
Iranfar et al. (2026) used dual-vector in mice. It worked in mice (perilymph 2.5 µL, tiny volume). The scaling to humans (191 µL) makes dual-vector dramatically less efficient. The single-vector solution doesn’t just improve efficiency — it changes the clinical feasibility calculus for human OHCs.
Connections
- STRC Mini-STRC Single-Vector Hypothesis — mini-STRC is the single-vector construct
- STRC AlphaFold3 Computational Experiments — Jobs 2, 5 validate mini-STRC
[see-also]STRC Anti-AAV Immune Response Model — single dose must succeed; immune barrier[see-also]Sonogenetic STRC Computational Proof — sonogenetic uses single-vector construct[see-also]Alternative STRC Delivery Hypotheses — sonoporation avoids dual-vector problem entirely[applies]STRC Hearing Loss — quantifies the improvement from mini-STRC[about]Misha- Misha-Hearing-10-Year-Plan
[source]2026-04-16-lesperance-otoferlin-gene-therapy — OTOF clinical calibration data[source]tarchini-2021-strc-dual-vector-neonatal — published dual-AAV STRC mouse data