Learn R Programming

pauwels2014 (version 1.0)

Bayesian Experimental Design for Systems Biology.

Description

Implementation of a Bayesian active learning strategy to carry out sequential experimental design in the context of biochemical network kinetic parameter estimation. This package gathers functions and pre-computed data sets to reproduce results presented in Pauwels E. et. al published in BMC Systems Biology, 2014. Scripts are given to compute all results from scratch or to draw pictures based on pre-computed data sets.

Copy Link

Version

Install

install.packages('pauwels2014')

Monthly Downloads

6

Version

1.0

License

GPL-3

Maintainer

Edouard Pauwels

Last Published

August 23rd, 2014

Functions in pauwels2014 (1.0)

sample_function

Generates posterior samples
risk_theta_fun

Risk function
simulate_experiment

Simulates the dynamics of a molecular perturbation
dream6_design

Simulates the active design process using the comparison criterion (see article for details).
exps

List of possible experiments
read_knobjs

Summarizes pre-computed results.
estimate_risk_dream6

Expected risk estimation (comparison with litterature).
pauwels2014-package

Reproduce numerical experiments
active_design

Simulates the active design process.
observables

Observable quantities of the model
generate_our_knowledge

Initialize a knowledge list.
transform_params

User defined parameter transformation function.
log_prior

User defined log prior
risk_theta_vect

Expected risk based on a posterior sample
log_likelihood

User defined likelihood function.
add_noise

Noise generative process for the simulations
generate_sample

An implementation of the Metropolis Hasting algorithm
simulate_experiment_no_transform

Link to the ode solver.
eval_log_like_knobj

Posterior function.
knobjs

Knowledge lists
random_design

Simulates a randim design process.
estimate_risk_out_all

Expected risk estimation.
log_normalize

Normalize in log space
sample_function_multi_mod_weight

Sample function visiting multiple modes of the posterior
experiment_list1

Molecular perturbations.
sample_function_single_mod

Sample function visiting a single mode of the posterior.
BFGS_special

An implementation of BFGS method for posterior maximization.
eval_kn_log_like

Evaluates the likelihood of a parameter value
proj_grad

Least square on the positive orthant
compute_gradient

Finite difference function
compute_mean_risks

Compute an average risk as a function of credit spent
armijo

Performs armijo line searcc
compute_gradient_coordinate

Finite difference function
dmvnorm

Gaussian multivariate density
add_infinitesimal

Finite difference function
reverse_params

Transform log space parameters back to the original space