qAOP Framework

This repository provides resources to generate figures and re-run model calibration for the study:
Developing Quantitative Adverse Outcome Pathways: An Ordinary Differential Equation-Based Computational Framework.
Published in Computational Toxicology. DOI: 10.1016/j.comtox.2024.100330.

Folder Contents

feedback_qAOP

This folder contains files for generating Figures 1 and 5:

• Figure 1: Run the code chunks in feedback_qAOP.Rmd. The bifurcation plot (Figure 1B) generation entails the call of grindnew.R for phase portrait and bifurcation analyses.
• Figure 5: Run the Python script sAOP_dist_panel.py. This script generates Figure 5 and saves individual and composite distribution plots in a timestamped folder within the simpleAOP directory.

model_comparison

This folder contains files for generating Figure 2:

• Run the R script model_comp.R. This script:

  • Loads the three qAOP model functions.
  • Generates artificial data for the selected model.
  • Prompts the user to choose two models for comparison by entering their corresponding capital letters (e.g., A, B, or C).
  • Prompts the user to select the model from which artificial data should be generated.
  • Performs cmdstanr parameter inference for both models using the generated artificial data.

• Calibration files (posterior draws, parameter summaries, LOO diagnostics, trace plots, and density plots) will be saved in a timestamped folder within the model_comparison directory.

• Model calibration is performed using files named qAOPX.stan, where X corresponds to the model selected (A, B, or C).

• To plot the model fits, run the code chunks in timeplots.Rmd.

model_updating

This folder contains files for generating Figure 3:

• Run all the code chunks in model_updating.Rmd to:
  • Define qAOP models.
  • Generate artificial data.
  • Run cmdstanr model fitting.

response-ODE

This folder contains files for generating Figure 4:

• Run all the code chunks in response-ODE.Rmd.