getLBodeContObjFunction: Returns the objective function to perform parameter...

Description Usage Arguments Details Value Author(s) See Also Examples

Description

This function configures returns the objective function that can be used to evaluate the fitness of a logic based ODE model using a particular set of parameters. This function can be particularly useful if you are planing to couple a nonlinear optimization solver. The returned value of the objective function corresponds to the mean squared value normalized by the number of data points.

Usage

1
2
3
4
	getLBodeContObjFunction(cnolist, model, ode_parameters, indices=NULL, time = 1, 
	verbose = 0, transfer_function = 3, reltol = 1e-04, atol = 0.001, maxStepSize = Inf, 
	maxNumSteps = 1e+05, maxErrTestsFails = 50, nan_fac = 1, lambda_tau=0, lambda_k=0,
	bootstrap=F, SSpenalty_fac=0, SScontrolPenalty_fac=0, boot_seed=sample(1:10000,1))

Arguments

cnolist

A list containing the experimental design and data.

model

The logic model to be simulated.

ode_parameters

A list with the ODEs parameter information. Obtained with createLBodeContPars.

indices

Indices to map data in the model. Obtained with indexFinder function from CellNOptR.

time

An integer with the index of the time point to start the simulation. Default is 1.

verbose

A logical value that triggers a set of comments.

transfer_function

The type of used transfer. Use 1 for no transfer function, 2 for Hill function and 3 for normalized Hill function.

reltol

Relative Tolerance for numerical integration.

atol

Absolute tolerance for numerical integration.

maxStepSize

The maximum step size allowed to ODE solver.

maxNumSteps

The maximum number of internal steps between two points being sampled before the solver fails.

maxErrTestsFails

Specifies the maximum number of error test failures permitted in attempting one step.

nan_fac

A penalty for each data point the model is not able to simulate. We recommend higher than 0 and smaller that 1.

lambda_tau

Tunable regularisation parameters to penalise L1-norm of parameters tau and induce sparsity. We recommend testing values between 0 and 100 (in log scale) to find best compromise between good fit and sparse model. Default =0, corresponding to no regularisation.

lambda_k

Tunable regularisation parameters to penalise L1-norm of parameters k and induce sparsity. We recommend testing values between 0 and 100 (in log scale) to find best compromise between good fit and sparse model. Default =0, corresponding to no regularisation.

bootstrap

If set to TRUE performs random sampling with replacement of the measurements used in the optimisation (to be run multiple times to get bootstrapped distribution of parameters). Default =FALSE, no bootstrapping.

SSpenalty_fac

Penalty factor for penalising solutions which do not reach steady state. Default =0.

SScontrolPenalty_fac

Penalty factor for penalising solutions for which the control (unperturbed) condition (assumed to be first row) does not reach steady state. Default =0.

boot_seed

Seed used for random sampling if bootstrap=TRUE. Default chose random seed between 0 and 10000

Details

Check CellNOptR for details about the cnolist and the model format. For more details in the configuration of the ODE solver check the CVODES manual.

Value

Returns a function to evaluate the model fitness. This function receives a vector containing both continuous parameters and integer values representing which reactions should be kept in the model.

Author(s)

David Henriques, Thomas Cokelaer, Federica Eduati

See Also

CellNOptR createLBodeContPars

Examples

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
 	library(CNORode)
	data("ToyCNOlist",package="CNORode");
	data("ToyModel",package="CNORode");
	data("ToyIndices",package="CNORode");
	
	ode_parameters=createLBodeContPars(model,random=TRUE);
	minlp_obj_function=getLBodeContObjFunction(cnolistCNORodeExample, model,ode_parameters,indices);
	
	x=ode_parameters$parValues;
	
	f=minlp_obj_function(x);

CNORode documentation built on Nov. 8, 2020, 7:39 p.m.