Modeling the homeostatic expansion of the immune system in a healthy and autoimmune system

Jonathan Anzules, Kristen Valentine, Genevieve Mullins, Katrina K. Hoyer, Suzanne Sindi

This project explores the systemic differences between a healthy and autoimmune system. We represent the homeostatic dynamics that work properly during healthy development and the cascading failure seen in the IL-2 knock system that leads to autoimmune disease. Below is a diagram of the model:

---

Where to Start

The QuickSimulation.m script is the easiest bit of code to go in and start exploring the dynamics of the model. Improvements to the accessibility of this script is planned for the future.

Below is a file tree of the most important parts of my project:

HomeostaticExpaction
|...Modeling
|    |...Matlab
|    |    |...CodesFrom_Khailaie2013
|    |    |...Matlab
|    |        |...RawData
|    |        |...Residuals
|    |        |...Examples
|    |        |...Plots
|    |        |...WorkingScript
|    |            |...ModelandCellGrowth.m
|    |            |...QuickSimulation.m
|    |            |...LHSInitialConditions.m
|    |            |...LHSParameters.m
|    |        |...Data
|    |...Python (Python Version, WIP by Pheobe Adamyan)
|
|...Stats plots and data management
    |...popCount_V2.R
    |...CalculatingActivatedTCellsFromCD44.py
    |...poCount_V2_AfterPythonScript.R
    |...StatsForPopCount_V2Results.Rmd

Defining Folders and Scripts

1. Modeling Folder - Contains the scripts that run the simulation, optimization/minimizing algorithm, and sensitivity analysis. Currently, only done in Matlab, but will also be available in Python in the future
2. Matlab Folder - Simulations and exploratory scripts written in Matlab
   • RawData Folder - Contains the data that my model attempts to replicate. Analyzed and prepared by scripts in the folder "Stats plots and data management"
   • Residuals Folder - Calculates the residuals between the model and data for any given parameter sets.
   • Examples - Various scripts that I used as inspiration for the structure of my algorithms.
   • WorkingScript - Contains all of the scripts necessary to run the simulation, fit model to data, and conduct sensitivity analysis
     • ModelandCellGrowth.m - The main script for minimizing the Rsquare value calculated by comparing the the simulation to the data.
     • QuickSimulation.m - Runs a simulation with the chosen parameter set and time frame. Plots are created for each run, if so desired. Individual values for each parameters can be edited to study the dynamics of the model
     • LHSInitialConditions - Performs a Latin hypercube sampling of any given population initial conditions and range. The 90th/10th percentile and one standard deviation above and below the mean is calculated. Plots displaying the area that the simulation results take is made for either percentiles or percentages.
     • LHSParameters - Performs a Latin hypercube sampling of any given parameter and a chosen range. The 90th/10th percentile and one standard deviation above and below the mean is calculated. Plots displaying the area that the simulation results take is made for either percentiles or percentages.
   • Data - Contains the data related to minimization/optimization, the most interesting parameter sets, and notes
     • ObjectiveDataKO.csv, ObjectiveDataWT - Contains all the data used for the calculation of the best parameter set chosen by fmincon(). This was used to explore how the data was used to calculate the objective, used for sanity checks.
     • ParametersSets.csv - Collection of parameters the display the most interesting dynamics seen in the model and progression towards the best fit of the model to data.
3. Stats plots and data management - This is where all of the raw data generated by the LSR/FCS-Express is analyzed and prepared for fitting.
   • popCount_V2.R - Takes the event numbers found for each cellular population and estimates the total population size collected from the samples
   • CalculatingActivatedTCellsFromCD44.R - Groups all of the related experiments and calculates how many of the CD4 T cells have been activated and how many are naive T cells
   • poCount_V2_AfterPythonScript.R - Finalizes the preparation of the data for model fitting
   • StatsForPopCount_V2Results.R - Calculates the statistical comparison of all cellular populations at different time points. Plots are also made for visualization of the data
4. Python - Code in this folder will replicate all simulations, optimization, and sensitivity analysis done in the Matlab folder, but in Python