Skip to contents

Get Started

Source: vignettes/esqlabsR.Rmd
esqlabsR.Rmd

Introduction

esqlabsR is designed to make life easier for OSPS users. The workflows implemented in this package rely on simple concepts:

  • a predefined project folder structure,
  • Excel files for definition of simulation scenarios out outputs.

Initialize Project

esqlabsR facilitates modeling and simulation workflows with OSPS relying on a well defined project structure based on Excel files. These files are organized in different folders and their structure must not be altered.

A new project folder can be initialized with the init_project() function.

It will create the expected simulation project’s folder structure, which is described in details in vignette("esqlabsR-project-structure").

Each one of these folders contains a series of .xlsx files with specific purpose. They are all linked to the project through the ProjectConfiguration.xlsx file located in the root of the project folder.

This file defines where all the necessary files are stored in the project folder. This file is already preconfigured and should not be changed in most cases. However, if target files would come to change location, ProjectConfiguration.xlsx should reflect that. All the path specified in the Value column should be relative to the ProjectConfiguration.xlsx location.

Loading the project configuration is the first step in any workflow using esqlabsR.

my_project_configuration <- createDefaultProjectConfiguration(path = "path/to/ProjectConfiguration.xlsx")
ProjectConfiguration: 
Relative path from working directory: C:/[path]/TestProject
Project Configuration File: Code/ProjectConfiguration.xlsx 
Model folder: Models/Simulations 
Parameters folder: Parameters 
Parameters file name: Parameters/ModelParameters.xlsx 
Individual parameters file name: Parameters/Individuals.xlsx 
Population parameters file name: Parameters/PopulationParameters.xlsx
Scenario definitions file name: Parameters/Scenarios.xlsx
...

Note that the ProjectConfiguration stores paths to other files that are used in the project.

To create an example ProjectConfiguration and execute the rest of this tutorial, run the following:

Design Scenarios

Now that the project’s file structure is ready, the next step is to run simulations. In esqlabsR, simulation are run by defining and executing multiple scenarios.

To modify, add or delete a scenario, edit the Parameters/Scenarios.xslx file. Each row of this file defines a simulation scenario.

The main properties that can be defined for a scenario are:

  • ModelFile is the name of the .pkml file that can be found in the Models/ directory.
  • ModelParameterSheets are the names of the sheets in Parameters/ModelParameters.xlsx that contain the parameters to apply to the model,
  • ApplicationProtocol is the name of the sheet in Parameters/ApplicationParameters.xlsx that specify the simulated administration protocol,
  • SimulationTime and SimulationTimeUnit define the time range to simulate, it should be three numbers for {start, end, resolution}
  • Individuald: the id of the individual to simulate. The id and settings for individuals are defined in Parameters/Individuals.xlsx,
  • PopulationId: the id of the population to simulate. The id and settings for populations are defined in Parameters/PopulationParameters.xlsx.

Note that multiple aspects of a simulation scenario are defined in other excel files, and the Scenarios.xslx file links to the specific sheets or other information stored in those files.

For example, if we want to change the characteristic of the individual used in the simulation, we must add a new row with a different IndividualId in the IndividualBiometrics sheet of theIndividuals.xlsx file and/or specify a parameter sheet with this IndividualId in the same file.

For more information on all the scenario parameters that can be setup, read vignette("esqlabsR-design-scenarios").

Run Scenarios

Once all the scenario properties are set up in the excel files, we can import ScenarioConfiguration using readScenarioConfigurationFromExcel() and create Scenario objects with createScenarios():

my_scenarios <- createScenarios(
  readScenarioConfigurationFromExcel(
    scenarioNames = "TestScenario",
    projectConfiguration = my_project_configuration
  )
)
#> ScenarioConfiguration: 
#>    Model file name: Aciclovir.pkml 
#>    Scenario name: TestScenario 
#>    Parameters sheets: Global 
#>    Individual Id: Indiv1 
#>    Population Id: NULL 
#>    Read population from csv file: FALSE 
#>    Application protocol: Aciclovir_iv_250mg 
#>    Simulation time intervals: NULL 
#>      Interval: 1 
#>        Start: 0 
#>        End: 24 
#>        Resolution: 60 
#>    Simulation time intervals unit: h 
#>    Simulate steady-state: FALSE 
#>    Steady-state time: 1000

Then, we run the simulations by passing the scenarios we defined:

myScenarioResults <- runScenarios(my_scenarios)

Simulation results can be saved for later use: saveScenarioResults() and loadScenarioResults().

saveScenarioResults(myScenarioResults,
  projectConfiguration = my_project_configuration,
  outputFolder = my_project_configuration$outputFolder
)

myScenarioResults <- loadScenarioResults(
  scenarioNames = "TestScenario",
  resultsFolder = my_project_configuration$outputFolder
)

Learn more on how to run simulations in vignette("esqlabsR-run-simulations").

Plot Results

The simulation results can now be plotted. For this, the package relies on on the concept of combining multiple simulation results and observed data using the DataCombined class and creating figures using functions implemented in the ospsuite package.

First, a DataCombined object is initialized:

my_datacombined <- DataCombined$new()

Then, simulation result are added to the dataCombined object:

my_datacombined$addSimulationResults(myScenarioResults$TestScenario$results,
  names = "Simulated",
  groups = "Aciclovir"
)

Finally, the plot is generated:

plotIndividualTimeProfile(my_datacombined)

In some cases, we also want to plot the observed experimental data (stored in Data/). DataCombined also has the ability to store observed data for them to be plotted.

observed_data <- loadObservedData(
  projectConfiguration = my_project_configuration,
  sheets = "Laskin 1982.Group A"
)

my_datacombined$addDataSets(observed_data, names = "Observed", groups = "Aciclovir")

plotObservedVsSimulated(my_datacombined)

Many other plot types are available in esqlabsR, read vignette("esqlabsR-plot-results") to learn more.