The construct
builds itself.
From disease name to optimized AAV vector in 6 deterministic steps. No black boxes. Every decision traceable to a primary reference.
Design Pipeline
Disease Input
Natural language disease query → normalized ID lookup with fuzzy matching
Gene Resolution
Primary gene identification, modifier genes, associated therapeutic targets
Target Ranking
Score genes by role (primary/modifier), AAV compatibility, and clinical significance
Tissue Mapping
Disease → target tissues → optimal promoter + serotype + route selection
Construct Assembly
Component-level sizing: ITRs + promoter + gene + enhancer + polyA + immunomod
Simulation
PK/PD modeling → expression kinetics → immune response → therapeutic window
Disease Explorer
Select a disease to see its gene targets and affected tissues. Live from chaos.db.
Gene Database
0 genes with exact coding sizes from primary references.
| Gene | Full Name | Locus | Coding (bp) | Mini-Gene | Significance |
|---|
Tissue Decision Matrix
Priority-ranked selections from 0 promoter and 0 serotype mappings.
| Tissue | Promoter | Size | Serotype |
|---|
Construct Templates
0 pre-computed reference designs.
Tool Modules
Three Python modules. Ten tool functions. Claude is the reasoning engine — tools provide data.
Gene info with coding size, chromosomal location, mini-gene availability
Disease profile with primary gene, inheritance, target tissues, related genes
Score and rank all genes for a disease by therapeutic potential
Gene vs 4,700bp AAV limit with mini-gene and dual-AAV strategies
Priority-ranked tissue → promoter from 23-row decision table
Full construct assembly with auto-sizing, fallbacks, capacity validation
Day-by-day PK/PD + immune + NAb simulation for any combination
Run simulation from a named construct template
Head-to-head comparison of all 6 serotypes
Data Layer
0 genes, 0 diseases, 0 regulatory elements, 0 templates loaded.