Stochastic approximation of signalling pathway dynamics
CREATED: 200809290400 Speaker: Bing Liu ** Motivation
- reactions in pathway described using ODE
- solving ODE needs to use small time step, time consuming
- Model analyses require a large number of simulations ** parameter estimation ** global sensitivity analysis ** perturbation analysis
- Biological systems are noisy
- Input data are coarse ** Approach using PGM
- approximate deterministic ODE using PGM
- pose questions as inference problems
- use discrete numbers to represent intervals
- a trajectory is sequence of states
- the dynamics is the set of all possible trajectories $\rightarrow$ state transition graph
- state transition graph $\rightarrow$ Dynamic Bayesian Network
- model analysis $\rightarrow$ Bayesian inference