Bayesian Reconstruction of Disease Outbreaks by Combining Epidemiologic and Genomic Data • outbreaker2

outbreaker2: a framework for reconstructing disease outbreaks

Welcome to the project page of outbreaker2, a Bayesian framework for integrating epidemiological and genetic data to reconstruct transmission trees of densely sampled outbreaks. It re-implements, generalises and replaces the model of outbreaker, and uses a modular approach which enables fine customisation of priors, likelihoods and parameter movements (see customisation vignette).

NOTE: Correction to genetic likelihood

The genetic likelihood of the original outbreaker paper was found to contain a minor mistake in accounting for unobserved generations of infection. As of June 7th 2019, outbreaker2 will use the correct genetic likelihood published here.

The original genetic likelihood was:

$\mu^{d(s_i,s_{\alpha_i})}(1 - \mu)^{(\kappa_i\times l(s_i, s_{\alpha_i})) - d(s_i,s_{\alpha_i})}$

The corrected genetic likelihood is:

$(\kappa_i \mu)^{d(s_i,s_{\alpha_i})}(1 - \mu)^{(\kappa_i\times l(s_i, s_{\alpha_i})) - d(s_i,s_{\alpha_i})}$

Installation

To install the stable version from CRAN:

install.packages("outbreaker2")

To install the development version from github (requires Rtools on windows and GSL headers on all platforms):

devtools::install_github("reconhub/outbreaker2")

To add local copies of the vignettes, you will need to specify:

devtools::install_github("reconhub/outbreaker2", build_vignettes = TRUE)

Then, to load the package, use:

library("outbreaker2")

Documentation

outbreaker2 is fully documented on a dedicated website.

It also comes with the following vignettes:

introduction: general introduction using a worked example.
overview: brief overview of the package’s content.
customisation: customisation of priors, likelihoods, and movement functions.
Rcpp_API: documentation for the Rcpp API.