"Stepping Back to School" report repository

This repository contains the code originally used for IDM's school reopening analysis presented in the Stepping Back to School modeling report, which is based on data from King County, WA, but explores fundamental COVID-19 transmission relationships that are broadly applicable. The report used the agent-based model Covasim, which can be downloaded from GitHub and used for other COVID-19 disease modeling.

• Code to implement schools in Covasim can be found in covasim_schools.
• The controller is implemented in covasim_controller.
• Other utility functions for running school analyses can be found in school_tools.
• Scripts to conduct the analysis are in scripts.
• Unit and integration tests are in tests.

Installation

Requirements

Python 3.7 or 3.8 (64-bit). (Note: Python 2 is not supported, Python <=3.6 requires special installation options, and Python 3.9 is not supported.)

Steps

1. If desired, create a virtual environment.

   • For example, using conda:

     conda create -n covaschool python=3.8
     conda activate covaschool
     

2. Install SynthPops, a package to create synthetic populations, in a folder of your choice. Note that pip installation does not currently include required Seattle data files:

   git clone https://github.com/InstituteforDiseaseModeling/synthpops
   cd synthpops
   pip install -e .
   

3. Install this package (which will also install Covasim):

   git clone https://github.com/InstituteforDiseaseModeling/stepping-back-to-school
   cd stepping-back-to-school
   pip install -e .
   

   If you want to install with webapp support (which is provided in a separate repository), use pip install -e .[web].

Usage

Scripts in the scripts folder produce the results presented in the report. Each script generates a different set of results; not all are used in the report. Each script has a brief description of what it does. For a quick example, first run python create_pops.py, then run python run_debug.py --show (the --show argument makes the plots appear when running non-interactively). This script should take a few minutes to run.

For more information, see the documentation in the individual files.