Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →AI-powered adeno-associated virus (AAV) vector design for gene therapy including capsid engineering, promoter selection, and tropism optimization.
.claude/skills/aav-vector-design-agent/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-13 | ✗→✓ | ▲ Improved | 23% | 0% |
| case-21 | ✗→✓ | ▲ Improved | 601% | 0% |
| case-01 | ✓→✓ | ＝ Same ✓ | 158% | 0% |
| case-02 | ✓→✓ | ＝ Same ✓ | 359% | 0% |
| case-03 | ✓→✓ | ＝ Same ✓ | 379% | 0% |
<!--
#
#
-->
The AAV Vector Design Agent provides AI-driven design of adeno-associated virus vectors for gene therapy applications. It covers capsid selection and engineering, promoter/enhancer design, transgene optimization, and manufacturing considerations.
| Serotype | Primary Tropism | Clinical Use | |----------|-----------------|--------------| | AAV1 | Muscle, CNS | Glybera (muscle) | | AAV2 | Broad (liver, muscle) | Luxturna (retina) | | AAV5 | CNS, liver, retina | Hemgenix (liver) | | AAV8 | Liver, muscle | Multiple trials | | AAV9 | CNS, cardiac, liver | Zolgensma (CNS) | | AAVrh10 | CNS, liver | CNS trials | | AAVrh74 | Muscle | Elevidys (muscle) | | AAV-PHP.eB | CNS (mouse) | Research |
User: "Design an AAV vector for liver-directed gene therapy in hemophilia B with low immunogenicity."
Agent Action:
bashpython3 Skills/Gene_Therapy/AAV_Vector_Design_Agent/aav_designer.py \ --target_tissue liver \ --therapeutic_gene F9 \ --indication hemophilia_b \ --minimize_immunogenicity true \ --nab_escape true \ --promoter liver_specific \ --output aav_design/
5' ITR - [Promoter] - [5' UTR] - [Transgene] - [WPRE] - [PolyA] - 3' ITR
Packaging limit: ~4.7 kb between ITRsPromoter Options: | Promoter | Type | Size | Application | |----------|------|------|-------------| | CAG | Ubiquitous | 1.7 kb | Strong expression | | EF1α | Ubiquitous | 1.2 kb | Constitutive | | LP1 | Liver-specific | 0.5 kb | Hepatocyte targeting | | hSyn | Neuron-specific | 0.5 kb | CNS applications | | MCK | Muscle-specific | 0.6 kb | Myopathies | | CMV | Ubiquitous | 0.6 kb | High initial (silenced) |
Directed Evolution:
Rational Design:
Machine Learning:
Pre-existing NAbs: | Serotype | NAb Prevalence | |----------|----------------| | AAV2 | 30-60% | | AAV5 | 15-30% | | AAV8 | 15-25% | | AAV9 | 20-35% |
Mitigation Strategies:
Tropism Prediction:
Immunogenicity Modeling:
Expression Optimization:
| Factor | Impact | Optimization | |--------|--------|--------------| | Capsid yield | Production cost | Sequence modifications | | Empty/full ratio | Potency | Purification method | | Aggregation | Stability | Formulation | | DNA packaging | Transgene size | Cassette design |
AI Group - Biomedical AI Platform
<!-- AUTHOR_SIGNATURE: 9a7f3c2e-MD-BABU-MIA-2026-MSSM-SECURE -->
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | pass→pass | 4,990 | 5,244 | +5% | 1 | 1 | 0% | 880 | 2,274 | +158% | 0 | 0 | — |
case-02 | pass→pass | 5,595 | 3,198 | -43% | 1 | 1 | 0% | 421 | 1,934 | +359% | 0 | 0 | — |
case-03 | pass→pass | 2,380 | 2,113 | -11% | 1 | 1 | 0% | 373 | 1,788 | +379% | 0 | 0 | — |
case-04 | pass→pass | 3,088 | 3,042 | -1% | 1 | 1 | 0% | 488 | 1,940 | +298% | 0 | 0 | — |
case-05 | pass→pass | 2,528 | 2,212 | -13% | 1 | 1 | 0% | 367 | 1,847 | +403% | 0 | 0 | — |
case-06 | pass→pass | 7,622 | 4,918 | -35% | 1 | 1 | 0% | 1,435 | 2,254 | +57% | 0 | 0 | — |
case-07 | pass→pass | 9,314 | 8,166 | -12% | 1 | 1 | 0% | 1,595 | 2,869 | +80% | 0 | 0 | — |
case-08 | pass→pass | 8,944 | 7,208 | -19% | 1 | 1 | 0% | 1,553 | 2,702 | +74% | 0 | 0 | — |
case-09 | pass→pass | 20,654 | 4,317 | -79% | 1 | 1 | 0% | 4,140 | 2,175 | -47% | 0 | 0 | — |
case-10 | pass→pass | 10,673 | 8,427 | -21% | 1 | 1 | 0% | 1,775 | 2,863 | +61% | 0 | 0 | — |
case-11 | pass→pass | 8,358 | 5,695 | -32% | 1 | 1 | 0% | 1,524 | 2,473 | +62% | 0 | 0 | — |
case-12 | pass→pass | 7,579 | 7,150 | -6% | 1 | 1 | 0% | 1,272 | 2,695 | +112% | 0 | 0 | — |
case-13 | fail→pass | 8,341 | 1,908 | -77% | 1 | 1 | 0% | 1,390 | 1,713 | +23% | 0 | 0 | — |
case-14 | pass→pass | 9,991 | 3,337 | -67% | 1 | 1 | 0% | 1,587 | 2,003 | +26% | 0 | 0 | — |
case-15 | pass→pass | 7,883 | 8,488 | +8% | 1 | 1 | 0% | 1,350 | 2,820 | +109% | 0 | 0 | — |
case-16 | pass→pass | 2,901 | 6,773 | +133% | 1 | 1 | 0% | 415 | 1,810 | +336% | 0 | 0 | — |
case-17 | pass→pass | 7,448 | 2,594 | -65% | 1 | 1 | 0% | 1,315 | 1,823 | +39% | 0 | 0 | — |
case-18 | pass→pass | 4,924 | 6,435 | +31% | 1 | 1 | 0% | 829 | 2,508 | +203% | 0 | 0 | — |
case-19 | pass→pass | 6,392 | 4,643 | -27% | 1 | 1 | 0% | 1,077 | 2,212 | +105% | 0 | 0 | — |
case-20 | pass→pass | 14,708 | 20,830 | +42% | 1 | 1 | 0% | 2,201 | 4,852 | +120% | 0 | 0 | — |
case-21 | fail→pass | 32,375 | 12,075 | -63% | 1 | 1 | 0% | 563 | 3,948 | +601% | 0 | 0 | — |
case-22 | fail→fail | 46,130 | 16,909 | -63% | 1 | 1 | 0% | 1,409 | 4,759 | +238% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted, and 20 counted toward the lift figure. The other 2 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +9 percentage points is the difference between those two pass rates over the 20 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
| Model | Method | Date | Lift |
|---|---|---|---|
| gemini-3.6-flash | verified | 7/24/2026 | +14% |
Other measured skills in the registry, with their headline benchmark lift.