This R code provides an implementation of the method for parameter estimation in ODE systems for biopathway models, as described in the paper "Inference in nonlinear differential equations" by M. Niu, M. Filippone, D. Husmeier, & S. Rogers, published in the proceedings of the 30th International Workshop on Statistical Modelling (IWSM), Linz, Austria, 06-10 Jul 2015, pp. 187-190:
http://eprints.gla.ac.uk/108342/
In order to run the R code, just type 'source('runloop3step.r')'. The parameter estimation results will be written out to the folder 'res', for which an example is provided with the software.
The default settings are for estimating all six parameters of the biopathway model described in the paper, using the MLP kernel for gradient matching. To change the noise level or the size of the data set, edit the files 'genData.r' and 'runloop3step.r'.
Here is an overview of all the R functions included in the software package:
runloop3step.r load the ODE observation data and carry out the parameter estimation.
genData.r set up the biopathway ODE model and generate ODE solution data.
setlam.r optimize the penalty parameter of the RKHS norm.
itertest.r the main iterative optimization loop.
MLP.r cross validation and optimisation functions.