Skip to content

MetaWards disease metapopulation analysis and modelling software. Professional geographical SIR model with a flexible plugin architecture to support complex scenario modelling

License

Notifications You must be signed in to change notification settings

metawards/MetaWards

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

MetaWards

Build status PyPI version Downloads DOI DOI

Scientific Background

MetaWards implements a stochastic metapopulation model of disease transmission. It can scale from modelling local transmission up to full national- or international-scale metapopulation models.

Please follow the quick start guide to see how to quickly get up and running using MetaWards to model your own custom disease or metapopulation model.

It is was originally developed to support modelling of disease transmission in Great Britain. The complete model description and the original C code are described here;

  • "The role of routine versus random movements on the spread of disease in Great Britain", Leon Danon, Thomas House, Matt J. Keeling, Epidemics, December 2009, 1 (4), 250-258; DOI:10.1016/j.epidem.2009.11.002

  • "Individual identity and movement networks for disease metapopulations", Matt J. Keeling, Leon Danon, Matthew C. Vernon, Thomas A. House Proceedings of the National Academy of Sciences, May 2010, 107 (19) 8866-8870; DOI:10.1073/pnas.1000416107

In this model, the population is divided into electoral wards. Disease transmission between wards occurs via the daily movement of individuals. For each ward, individuals contribute to the force of infection (FOI) in their home ward during the night, and their work ward during the day.

This model was recently adapted to model CoVID-19 transmission in England and Wales, with result of the original C code published here;

  • "A spatial model of CoVID-19 transmission in England and Wales: early spread and peak timing", Leon Danon, Ellen Brooks-Pollock, Mick Bailey, Matt J Keeling, Philosophical Transactions of the Royal Society B, 376(1829); DOI:10.1098/rstb.2020.0272

This Python code is a port which can identically reproduce the outputs from the original C code as used in that work. This Python code has been optimised and parallelised, with additional testing added to ensure that development and scale-up of MetaWards has been robustly and efficiently conducted.

Program Info

The package makes heavy use of cython which is used with OpenMP to compile bottleneck parts of the code to parallelised C. This enables this Python port to run at approximately the same speed as the original C program on one core, and to run several times faster across multiple cores.

The program compiles on any system that has a working C compiler that supports OpenMP, and a working Python >= 3.7. This include X86-64 and ARM64 servers.

The software supports running over a cluster using MPI (via mpi4py) or via simple networking (via scoop).

Full instructions on how to use the program, plus example job submission scripts can be found on the project website.

Installation

Full installation instructions are here.

Binary packages are uploaded to pypi for Windows, OS X and Linux (manylinux). The easiest way to install is to type in the console:

pip install metawards

(this assumes that you have pip installed and are using Python 3.7 or above - if this doesn't work please follow the full installation instructions).

Alternatively, you can also install from within R (or RStudio) by typing;

library(devtools)
install_github("metawards/rpkg")
metawards::py_install_metawards()

But, as you are here, I guess you want to install the latest code from GitHub ;-)

To do that, first clone and install the requirements;

git clone https://github.com/metawards/MetaWards
cd MetaWards
pip install -r requirements.txt
pip install -r requirements-dev.txt

Next, you can make using the standard Python setup.py script route.

CYTHONIZE=1 python setup.py build
CYTHONIZE=1 python setup.py install

Alternatively, you can also use the makefile, e.g.

make
make install

(assuming that python is version 3.7 or above)

You can run tests using pytest, e.g.

METAWARDSDATA="/path/to/MetaWardsData" pytest tests

or you can type

make test

You can generate the docs using

make doc

Running

You can either load and use the Python classes directly, or you can run the metawards front-end command line program that is automatically installed.

metawards --help

will print out all of the help for the program.

Running an ensemble

This program supports parallel running of an ensemble of jobs using multiprocessing for single-node jobs, and mpi4py or scoop for multi-node cluster jobs.

Note that mpi4py and scoop are not installed by default, so you will need to install them before you run on a cluster (e.g. pip install mpi4py or pip install scoop).

Full instructions for running on a cluster are here

History

This is a Python port of the MetaWards package originally written by Leon Danon. This port has been performed with Leon's support by the Bristol Research Software Engineering Group.

About

MetaWards disease metapopulation analysis and modelling software. Professional geographical SIR model with a flexible plugin architecture to support complex scenario modelling

Topics

Resources

License

Code of conduct

Security policy

Stars

Watchers

Forks

Languages

  • Python 81.4%
  • C 13.9%
  • Cython 4.4%
  • Other 0.3%