Blouch: Bayesian Linear Ornstein-Uhlenbeck models for Comparative Hypotheses fits allometric and adaptive models of continuous trait evolution in a Bayesian framework based on fixed or continuous predictors and incorporates measurement error. In addition to assigning biologically meaningful priors when compared to non-Bayesian approaches, Blouch includes new implementations of Ornstein-Ulenbeck models including allowing for varying effects (varying intercepts and varying slopes), multilevel modeling, and non-centered models.
While the front-end component of Blouch is written in R (R Core Team, 2023), the nuts and bolts are written in the language Stan (Carpenter et al., 2017), which allows estimation of Bayesian models using Markov chain Monte Carlo (MCMC) methods based on the Hamilton Monte Carlo sampler.
Getting Started
If you are just getting started with Blouch I recommend starting with the Empirical and Simulation Example articles available on the package website. The other articles are abbreviated versions of this example showing the various models implemented by Blouch - most of the steps of a preliminary analysis will be the same.
Blouch is based on an article:
- Grabowski, M (2024). Blouch: Bayesian Linear Ornstein-Uhlenbeck models for Comparative Hypotheses. Systematic Biology.
Instalation Instructions
To install the R and Stan functions associated with Blouch from github, first install the package devtools:
install.packages("devtools", repos = "https://cran.ma.imperial.ac.uk/")
library(devtools)Then install Blouch
devtools::install_github("mark-grabowski/blouch")
library(blouch)Documentation
For a complete walkthrough of the package visit the package website: here.