ttiq-simulation

Various models for assessment of test, trace, isolation and quarantine (TTIQ), used in modelling for National Cabinet in October/November 2021

Unfortunately it is not possible for anyone to run the code because it uses confidential health data. We aim to get approval to share summary or synthetic data for publication. More context is in the umbrella repository here.

Contributing to this repo?

If you are contributing to this repo, please create a branch or fork of the project and contribute via a pull request - this will avoid issues with everyone working on master at the same time.

Running analyses

Step 1: Install capsule

install.packages("capsule", repos = "https://milesmcbain.r-universe.dev")

or

remotes::install_github("milesmcbain/capsule")

Step 2: Reproduce the libraries used

capsule::reproduce_lib()

This recreates all of the R packages used in the analysis on your computer. Importantly, this will not change where your existing R packages are installed. It is just for this repository. So no need to be concerned about this impacting other analyses you run.

Step 3: Run the target workflow

capsule::run(targets::tar_make())

This runs our targets workflow using the R packages specified.

This will check if the targets are written, and if they aren't, it will re-run the necessary ones. In this case, the targets are stored in the _targets/objects folder.

Note that you will for most cases need access to data

Step 4 - ?

If you install some new packages or use a new package, add a library call to packages.R and then run:

create(dep_source_paths = "./packages.R")

To update the lockfile

Managing package conflicts

To make sure the right package function names are called, and avoid pain, for example, with dplyr::select, we use conflicted in the packages.R file.

This code:

conflict_prefer("filter", "dplyr")
conflict_prefer("select", "dplyr")

Means that filter and select will always be dplyr::filter and dplyr::select.

In addition, any conflicts that come up will cause an error.

Running tests

You can run regression tests against the images created by running:

testthat::test_dir("tests/testthat")

This will then compare the current plots in targets against a "golden standard" plots in _snaps.

Note that there might be some warnings - these are usually warnings from ggplot that say something like:

Removed 6 rows containing missing values (geom_col)

These are typically known cases where we have NA values in an experiment grid as there might not be valid values for a given combination of parameters.

The testing for these plots are created in tests/testthat/test-plots.R

To review existing plots against new ones, run:

testthat::snapshot_review("test-plots")

Note that the tests are not run as part of the targets workflow, as it is a separate step of the workflow.