EvoSim
Models how tumor subpopulations compete and evolve
Shannon entropy matches clinical samples
Matches multi-region sequencing patterns
No explosions in 10,000 runs
Phylogeny reconstruction accuracy
Overview
A stochastic simulation that models how different tumor subpopulations compete and evolve over time — which clone expands, which one gets suppressed, and when new resistant subclones emerge through mutation. It extends the deterministic tumor simulation with realistic biological randomness, producing not a single predicted trajectory but a range of possible outcomes with confidence intervals.
Inputs
4 inputstrajectory[T, K]Deterministic clone trajectories from Neural ODE
sigma[1]Stochastic noise scale from Hypernet
mutation_rate[1]Per-cell mutation probability
fitness_landscape[K, M]Fitness effects of M potential mutations
Outputs
4 outputsstochastic_trajectory[T, K']Clone populations with new subclones (K' ≥ K)
clone_treeGraphPhylogenetic tree of clone relationships
tumor_burden[T]Total tumor burden with stochastic effects
risk_score[1]Aggregate risk from evolutionary trajectory
Mathematical Formulation
Euler-Maruyama discretization with confidence cones (5th-95th percentile)
Multiplicative fitness from mutations
Poisson mutation process
Key Features
- Euler-Maruyama stochastic integration
- Mutation-driven subclone emergence
- Fitness-based selection pressure
- Clone extinction and emergence tracking
- Phylogenetic tree construction
Key Innovations
- 1Bridging deterministic ODE with stochastic evolution
- 2Realistic tumor heterogeneity modeling
- 3Treatment resistance emergence simulation
- 4Evolutionary risk scoring
Hyperparameters
0.01 (days)From Hypernet1e-6 per cell per day100 cells5010 cellsTraining Details
EvoSim is a simulation module, not trained. Parameters come from Hypernet (sigma) and configuration (mutation rates, fitness landscape). Validated against multi-region sequencing data.