MetStaT (version 1.0)

RevNet.JacobianMethod: Reverse engineering of networks: Penalized Jacobian method

Description

Network connections are estimated by calculating the Jacobian Matrix of the network. Details of algorithm and the theory behind the algorithm can be found in the references section. This method needs high frequency sampling data of small perturbations and steady state concentrations.

Usage

RevNet.JacobianMethod(data, delta.t, steady.state.concentrations, lamba.penalty.par, 
kappa.penalty.par, jacobian.threshold)

Arguments

data

multi-dimensional sampling matrix: time x variables x experiments

delta.t

time between subsequent time points used to calculate the fourth order approximation to the Jacobian. It the data are not evenly sampled interpolation can be used to obtain evenly distributed data with time interval delta.t.

steady.state.concentrations

a vector indicating the steady concentrations of the variables (must match second dimension size of 'data')

lamba.penalty.par

lambda penalty parameter to ensure sparsity in the Jabocian matrix

kappa.penalty.par

kappa penalty parameter to ensure sparsity in the Jabocian matrix

jacobian.threshold

threshold above which values in the found Jacobian matrix indicate an edge in the network.

Value

A connectivity matrix is returned.

References

Reverse engineering of metabolic networks, a critical assessment. Diana M. Hendrickx, Margriet M. W. B. Hendriks, Paul H. C. Eilers, Age K. Smilde and Huub C. J. Hoefsloot. Mol. BioSyst, Volume 7:2 (2011) pages 511-520

Examples

data(RevNetJacobian)
RevNet.JacobianMethod(RevNetJacobian, 0.01, SteadyState, 0.0002, 1, 0.0001)