generate_our_knowledge: Initialize a knowledge list.

Description Usage Arguments Details Value Author(s) Examples

View source: R/generate_our_knowledge.R

Description

This function defines a new knowledge list. Its main purpose is to give a value to all the global parameters that have to be set up before running the simulations.

Usage

1
generate_our_knowledge(transform_params, global_parameters, datas)

Arguments

transform_params

A function that transform parameter values. See transform_params.

global_parameters

A list of global parameters. See the details section for the default value.

datas

Potential data to be put in the datas slot.

Details

There are four main slots in the knowledge list the first one is transform_params which is by default a log transformation function that allows to work in log space. See transform_params. The second slot, global_parameters contains various global parameters necessary for the simulation to proceed. The default value of those parameters is given in the next section. See also the vignette for details. The third slot is named datas. It is empty by default, but is intended to receive datasets, experiment function, posterior sample and expected risk estimates for each step of the simulation. The vignette provides more details about this.

Value

A list that agregates the input. The default value of the global parameter is as follows:

atol = 1e-6

deSolve parameter (see deSolve)

rtol = 1e-6

deSolve parameter (see deSolve)

tspan = seq(0,100,0.5)

deSolve parameter, the time points at which all simulated kinetic quantities should be evaluated

max_step = 50

deSolve parameter (see deSolve)

max_it = 200

optimization parameter, the maximum number of iteration in the for BFGS_special function

tol = 1e-3

optimization parameter, a stoping criterion based on gradient norm tolerance for BFGS_special function

beta = 2

optimization parameter, the division to perform on the step size when armijo's rule fails in BFGS_special function

c = 0.0001

optimization parameter, armijo's rule second parameter for BFGS_special function

n_multi_mod_weight = 20

sampling parameter, number of times the local search/sample operations are repeated in sample_function_multi_mod_weight function

max_log_like = - 700

sampling parameter, a lower bound under which parameter values are systematically after the BFGS search in sample_function_multi_mod_weight function and in sample_function_single_mod function

centrality_ratio = 0.4

sampling parameter, allows to only keep parameter values leading to reasonable datafits by filtering those which lead to kinetics trajectories that do not pass in the "middle" of the data in sample_function_multi_mod_weight function and in sample_function_single_mod function

sample_burn_in = 5000

sampling parameter, size of the burn in sample to be ignored in the Metropolis Hasting algorithm algorithm in generate_sample

sample_to_keep1 = 10000

sampling parameter, sample size to be further sampled in MH algorithm in generate_sample

sample_step = 1

sampling parameter, the mean MH step length in generate_sample function

final_sample = 10000

sampling parameter, the final sample size provided by sampling functions sample_function_multi_mod_weight and sample_function_single_mod. This is the size of the sample that is to be stored in the knowledge list.

final_sample_design = 100

size of a subsample to be used for risk estimation in active_design strategy.

n_simu_weights = 100

number of noise simulations required to estimate the weights in risk estimation in active_design strategy.

initial_conditions = c(g6 = 1, p6 = 1,p7 = 1,p8 = 1, v6_mrna = 0,v7_mrna = 0,v8_mrna = 0)

simulation parameter, the initial condition default value

n_params = 9

simulation parameter, number of free parameters

param_names = c("p_degradation_rate", "r6_Kd", "r11_Kd", "pro6_strength", "pro7_strength", "pro9_strength", "rbs6_strength", "rbs7_strength", "rbs8_strength")

simulation parameters, the names the free parameters

params = c(p_degradation_rate = 1, r6_Kd = 1, r11_Kd=1, pro6_strength = 1, pro7_strength = 1, pro9_strength = 1, rbs6_strength = 1, rbs7_strength = 1, rbs8_strength = 1)

simulation parameters, an instance of free parameter named numeric vector

true_params = c(mrna6_degradation_rate =1, mrna7_degradation_rate =1, mrna8_degradation_rate =1, p_degradation_rate = 0.1, r6_Kd = 2.6, r6_h = 4, r11_Kd=2, r11_h = 2, r12_Kd = 0.2, r12_h = 2, pro6_strength = 1, pro7_strength = 0.8, pro9_strength = 3.77, rbs6_strength = 5, rbs7_strength = 5, rbs8_strength = 5)

simulation parameters, the true parameter to be estimated and used for simulations in the physical space

true_params_T = c(p_degradation_rate = 50, r6_Kd = 56.9162224661803, r11_Kd = 55.0171665943997, pro6_strength = 50, pro7_strength = 48.3848331165323, pro9_strength = 59.6056891700965, rbs6_strength = 61.649500072267, rbs7_strength = 61.649500072267, rbs8_strength = 61.649500072267)

simulation parameters, the true parameter to be estimated and used for simulations in log space, such that transform_params(true_params_T) = true_params

,

dllname = "pauwels2014"

simulation parameters, the name of the shared object which contains the function to be bassed to deSolve solver. See the vignette and deSolve package for more details. The default is "pauwels2014" as the source for the example of the article is provided in this package.

Author(s)

Edouard Pauwels

Examples

1
2

pauwels2014 documentation built on May 29, 2017, 9:03 a.m.