Stochastic modelling for quantitative description of heterogeneous biological systems

Cellular dynamics are intrinsically noisy, so mechanistic models must incorporate stochasticity if they are to adequately model experimental observations.As well as intrinsic stochasticity in gene expression, there are other sources of noise and heterogeneity in cells and cell populations.There is a well-developed framework for stochastic modelling, including algorithms for fast, approximate simulation of cellular dynamics.Multiscale models are particularly challenging, and are likely to require the use of fast stochastic emulators.Statistical modelling is concerned with relating models (either stochastic or deterministic) to experimental data, and as such is of key importance in systems biology.Simple statistical models are useful for fitting to high-throughput data such as time course microarray data for uncovering structural relationships between genes.The parameters of complex dynamic models can be estimated from high-resolution dynamic data using sophisticated statistical inference technology.A nonlinear multivariate stochastic differential equation model known as the chemical Langevin equation provides a natural bridge between simple structural statistical models and detailed mechanistic dynamic models.