Simulation Engine

Predict gene therapy
before you build it.

Osmium models expression kinetics, immune clearance, NAb interference, and therapeutic windows for AAV gene therapy constructs — from first principles, not black boxes.

14
Genes Modeled
7
Diseases Covered
6
AAV Serotypes
26
Regulatory Elements

Simulation Capabilities

Six modeling subsystems. One integrated prediction.

PK/PD

Expression Kinetics

Sigmoidal onset, exponential decay, CMV promoter silencing — modeled day-by-day for 365+ days per construct.

  • Three-phase model: zero → sigmoidal ramp → exponential decay
  • Cell division dilution based on tissue-specific turnover rates
  • CMV promoter silencing half-life at 90 days in vivo
  • Promoter strength multipliers for 11 different promoters
IMMUNOLOGY

Immune Response Modeling

Innate + adaptive immunity with tissue-specific immune privilege. Predicts T-cell clearance and its impact on expression.

  • Gaussian innate response peaks at day 3
  • Adaptive response onset at day 14, plateaus over weeks
  • Immune privilege factors: retina 0.9, brain 0.7, liver 0.0
  • Immunomodulator cassettes reduce response by 60%
IMMUNOGENICITY

NAb Seroprevalence

Pre-existing neutralizing antibodies block initial transduction. Population-stratified data (general vs pediatric) for all 6 serotypes.

  • AAV2 highest seroprevalence at 72% (general population)
  • Pediatric populations 40-50% lower across all serotypes
  • NAb reduction formula: max(0, 1 - prevalence × 1.5)
  • De novo NAb development modeled from day 14 onward
CLINICAL

Therapeutic Window Prediction

Automatically identifies when expression crosses the therapeutic threshold and when it falls below — the true clinical value.

  • Configurable therapeutic threshold per construct
  • Onset day, end day, and effective duration calculated
  • Redose recommendation 14 days before window closes
  • High NAb titer warning for same-serotype redosing
COMPARISON

Serotype Head-to-Head

Compare all 6 AAV serotypes against the same gene/tissue/route combination. Sorted by effective therapeutic duration.

  • Tissue-specific transduction efficiency multipliers
  • Route delivery efficiency (IV, intrathecal, subretinal, etc.)
  • Results ranked by effective duration, not just peak expression
  • Identifies optimal serotype for each tissue target
DOSING

Dose-Response Analysis

Model expression peaks across 6 orders of magnitude of dosing. Understand the dose-response relationship for each serotype.

  • Doses from 1e11 to 1e14 viral genomes
  • Peak expression scales with transduction efficiency
  • Tissue mass and blood flow fraction factored in
  • Ceiling effect at 100 AU prevents runaway expression

Roadmap

Where we are and where we're going.

PHASE 1FoundationSHIPPED
  • +SQLite database with 18 tables, 189 seed rows
  • +14 genes with exact coding sizes from primary references
  • +26 regulatory elements with bp-accurate sizing
  • +6 serotype tropism profiles + 8 administration routes
  • +Tissue-to-promoter and tissue-to-serotype decision tables
  • +117 passing tests with full data integrity verification
PHASE 2Simulation EngineSHIPPED
  • +PK/PD expression kinetics (onset → peak → decay)
  • +Immune response modeling (innate + adaptive)
  • +NAb seroprevalence-based transduction reduction
  • +12 tissue parameter presets with cell turnover data
  • +Serotype head-to-head comparison engine
  • +Therapeutic window detection and redose recommendation
PHASE 3Construct Design AgentSHIPPED
  • +AAV packaging compatibility check (4,700bp limit)
  • +Auto mini-gene fallback for oversized genes
  • +Tissue-optimized promoter selection (priority-ranked)
  • +Serotype and route recommendation engine
  • +Component-level size accounting (ITRs + promoter + gene + enhancer + polyA)
  • +6 pre-computed construct templates (SMA, LCA, HemB, SCD, CF, DMD)
PHASE 4External API IntegrationPLANNED
  • NCBI E-utilities for gene sequence validation
  • Ensembl REST API for transcript data
  • UniProt REST API for protein function
  • GTEx Portal for tissue expression data
  • gnomAD for population variant frequencies
  • bioRxiv/PubMed for literature monitoring

See it in action.

Explore the interactive visualizer with live simulation data from all 6 construct templates.

Launch Visualizer →