IDM Compartmental Modeling Software (CMS) in a Docker Container

Includes Python interface example using Python.NET

Quick Start with Docker

Get a local copy of the docker container image

Option 1- skip to examples below and let Docker pull the requisite image

Option 2- pull the image from Docker Hub:

docker pull clorton/idmcms:1.1

Option 3- build the image locally:

docker build --tag clorton/idmcms:1.1 .

Run a Model with CMS

Options:

Note: the examples below use Unix/Linux bash syntax to get the current directory, $(pwd). If running on Windows, explicitly list the current path in place of $(pwd).

1. Run the default model/command in the container (python3 /cms/seir.py).

docker run --rm -it -v $(pwd):/host -w /host clorton/idmcms:1.1

This will write trajectory data to trajectories.net.csv and a plot to trajectory.png in the current directory, mapped to /host in the container. The model is not modifiable as this default command runs the model included in the container. See options 2 or 3 for running other models.

2. _Run the model of your choice, written in EMODL, directly in CMS with the Mono .Net runtime.

docker run --rm -it -v $(pwd)/models:/host -w /host clorton/idmcms:1.1 mono /cms/compartments.exe --model seir.emodl --config seir.cfg

• Replace $(pwd)/models with your model directory path.
• Replace seir.emodl and seir.cfg with your model file and configuration, respectively.

This will write trajectory data to file specified in the configuration.

3. Run a model, e.g. SEIR, with Python.

docker run --rm -it -v $(pwd)/models:/host -w /host clorton/idmcms:1.1 python3 seir.py --png

This example, called from the root of the repository on your machine or codespace, runs the example seir.py, SEIR model implemented in Python and writes trajectories.net.csv and trajectory.png into the mapped directory (./models in the example).

Sample Output

SEIR Trajectory Data (CMS)

FrameworkVersion,"1.0.176.23","clorton (CMS/Main/framework)"
sampletimes,0,1,2,3,...,179,180
S{0},990,983,979,975,...,372,375
E{0},0,5,7,7,...,0,0
I{0},10,10,12,10,...,0,0
R{0},0,2,2,8,...,628,625
cumulative{0},0,7,11,15,...,938,938
population{0},1000,1000,1000,1000,...,1000,1000

Note that CMS can run multiple realizations of the model in a single invocation. The realization index is appended to each column to differentiate results from separate realizations.

SEIR Trajectory Data (Python)

,S{0},E{0},I{0},R{0},cumulative{0},population{0}
0,990.0,0.0,10.0,0.0,0.0,1000.0
1,985.0,4.0,9.0,2.0,5.0,1000.0
2,982.0,6.0,9.0,3.0,8.0,1000.0
3,977.0,9.0,7.0,7.0,13.0,1000.0
...
178,292.0,0.0,0.0,708.0,1040.0,1000.0
179,294.0,0.0,0.0,706.0,1040.0,1000.0

Note that Python trajectory data comes from a Pandas Dataframe and omits the "index" column name in the first column of the first row.

SEIR Trajectory Plot (Python)

Additional Information and Documentation

• PyCMS : The initial motivation for packaging CMS in a container.
• IDM-CMS : Source code and documentation for the CMS engine.