Parameter estimation of chemical reaction networks
MetadataShow full item record
This thesis describes a novel Monte Carlo simulation algorithm for the estimation of the model parameters of kinetic rate equation systems, describing biochemical reaction networks; and for the quantitative prediction of the time-dependent behavior of real biochemical systems described by such kinetics models. This simulation method, referred to as the super-ensemble approach, combines Monte Carlo sampling of the kinetics model parameter space with a simultaneous Galerkin-type variational Monte Carlo solution of the underlying kinetic rate equation system. Unlike the recently proposed and closely related “standard” ensemble simulation method, the super-ensemble does not rely on the high-volume execution of a conventional serial ordinary differential equation(ODE) solver algorithm, and it is therefore amenable to an efficient scalable parallelization by straightforward time domain decomposition techniques. With minor modifications, the super-ensemble algorithm can also be deployed as a parallelizable variational ODE solution method, in a conventional ODE solver setting where a unique ODE solution is sought for given initial conditions and given rate functions. Test applications of the super-ensemble algorithm in both ODE solver mode and in parameter estimation mode, for a simple enzyme catalysis model, will be discussed.