This repository contains scripts necessary to replicate the analyses of
Smith & Yang (2026), “Directional Biases in Morphological Evolution: Implications for Phylogenetic Models”.
Usage details are provided in the package vignettes.
Our analyses were conducted on the Hamilton HPC service at Durham University.
Each analysis (dataset × model) is conducted using an MPI installation of RevBayes.
Morpohological datasets are referred to using their MorphoBank project identifiers. Models are referred to using the name of their RevBayes implementation file, found in inst/rbScripts.
Once available, results are committed to a GitHub repository named by concatenating the dataset and model identifiers.
This R package then fetches the results via git, and caches a sub-sample of results locally. Analyses performed by the package are also cached.
Get started by checking out this repository from GitHub. Instructions are in the installation vignette.
Details of how to configure matrices for analysis are given in the matrix processing vignette.
A workflow to begin analysis can then be followed using ?EnqueueMC. Completed analyses can be retrieved from the server using ?Collect or ?UpdateRecords.
Further analyses are then detailed in the individual vignettes.
To reproduce our results from scratch, you will need to clear cached data. This includes:
inst/matrices: Matrices obtained from MorphoBank, and manual annotations. This directory is a git submodule linked to neo-trans/matrices
inst/data: Cached tree similarity measures (obtained using TreeSimilarities())
inst/projects: Nexus files corresponding to analysis partitions, produced using PrepareMatrix()
inst/results: Results of MCMCMC analysis; convergence diagnostics (-conv.txt); and marginal likelihood estimates (-stone.txt).
Analyses were queued using EnqueueML() and EnqueueMC(), with slurm configuration files in inst/slurm. Metadata from completed slurm jobs is recorded in inst/sacct.csv.