A direct derivative method for estimating kinetic parameters of biological networks

Challenged by strong nonlinearity of cellular network models, large uncertainty in model parameters, and noisy experimental data, a new parameter estimation algorithm, direct derivative method (DDM), is presented in which the measurement data are firstly fitted with smoothing splines, and then the first-order derivative of state variables are evaluated and substituted into the model. Thus, a dynamic optimization problem is converted into a linear or nonlinear regression problem. There is no need to solve ordinary differential equations of the system models iteratively, the computational complexity is therefore reduced to a large extent. Taking the IκBα-NF-κB signal transduction pathways as an example, unknown parameters are estimated effectively using the proposed DDM algorithm, and various factors that affect the results are investigated.