Access to PEP in Madagascar

This repository accompanies the manuscript:

Rajeev, M. et al. How access to care shapes disease burden: the current impact of post-exposure prophylaxis and potential for expanded access to prevent human rabies deaths in Madagascar. medrxiv preprint here

Code to reproduce the analyses and figures are in analysis/src. The Rmarkdown document and associated supporting files to reproduce the manuscript are in analysis/paper.

Reproducing results

All analyses were completed in R version 4.0.2. All rmarkdown outputs require rmarkdown.

Step 1: Clone the repository & also download intermediate outputs from zenodo.

git clone https://github.com/mrajeev08/MadaAccess.git

Also, download the outputs from this zenodo link(also includes some of the raw data used in the analysis, although not all due to sharing restrictions).

Step 2: Install dependencies

Option 1: using renv (will install version specific to renv & local package library) In R or Rstudio run this command:

renv::restore()

This will throw a warning if you're using a different R version from 4.0.2 used for this analysis.

Option 2: Source the script analysis/src/00_install.R (will install most recent to global package library).

You will also need to download and install JAGS.

Step 3: Run the analysis

The scripts are numbered sequentially according to how the should be run.

If you're in a unix environment, you can use the bash script runit.sh to run all analysis steps sequentially by:

./runit.sh -d "analysis/src/*/*" --printErrors -q

or to just make the figures:

./runit.sh -d "analysis/src/figures/*" --printErrors -q

To run the steps which were completed on a cluster, you will need to pass the -cl argument:

./runit.sh -d "analysis/src/*/*" --printErrors -q -cl

and then enter 1 to run the scripts in parallel or 2 to run serially on your local computer.

The easiest way to make the paper is in it's html format. To do so in R run:

rmarkdown::render(here("analysis/paper/manuscript.Rmd"), 
                  "bookdown::html_document2",
                  output_dir = here("analysis/paper"))

Some of these analysis steps require significant compute time & RAM, see the requirements here. If you do have ssh access to a remote cluster with a slurm scheduler and want to run jobs, I wrote this() shell utility to run the cluster jobs which you could adapt accordingly.

To see the help for runit.sh, run ./runit.sh -h on the command line.

Note that for the shell scripts, you may need to make them executable, i.e.:

chmod u+x runit.sh

Acknowledgements & hat-tips

Figure inspiration:
- Fig 3 inspiration from Gytis Dudas
- Fig 5 inspiration from Maarten Lambrechts
- Geom raincould gist from David Robinson
Resources on building a research compendium:
- rrtools
- workflowr
- renv

To do

Docker/binder
Pkgdown site