AuTuMN

This project is a modelling framework used by the AuTuMN tuberculosis modelling project. It provides a set of Python models that modularises the development of dynamic transmission models and allows a pluggable API to other system. Applied to tuberculosis.

The tuberculosis-specific AuTuMN modelling framework is build on top of the disease-agnostic SUMMER project.

See this guide for information on how to set up this project.

Project Structure

├── .github                 GitHub config
├── apps            Specific apps of AuTuMN
├── autumn                  AuTuMN framework module
├── data                    Data to be used by the models
├── docs                    Documentation
├── scripts                 Ad-hoc utility scripts
├── tests                   Automated tests
├── .gitignore              Files for Git to ignore
├── .pylintrc               PyLint code linter configuration
├── requirements.txt        Python library dependencies
└── setup.py                Packaging for deployment to MASSIVE computing platform

MASSIVE

We sometimes need to run jobs on Monash's MASSIVE computer cluster. The scripts and documentation that allow you to do this can be found in the scripts/massive/ folder.

Tests

Automated tests may be run via PyCharm or via the command line:

./scripts/test.ps1

Tests are also run automatically via GitHub Actions on any pull request or commit to the master branch.

Formatting

The codebase can be auto-formatted using Black:

./scripts/format.ps1

Running apps

Specific uses of the AuTuMN framework are present in apps/. You can run an application through an IDE like PyCharm, or run it from the command line:

./scripts/run.ps1 --help

Old notes below: are these used anymore?

TODO

document Bulgaria interventions properly in handbook
the model would not run without age-stratification (detected when running Bulgaria)

major outstanding tasks

mapping to DALYs, QALYs

minor tasks

simplify code for automatic detection of int_uncertainty start_time. Should use common method with optimisation start_dates
in the adjust_treatment_outcomes_support method, only the "relative" approach accounts for baseline intervention coverage The "absolute" approach should be updated similarly in case we use it with a non-zero coverage at baseline.

MattSegal / AuTuMN