TraitLab implements the stochastic Dollo model for simulating and fitting binary data on trees using Markov chain Monte Carlo methods.
- Original paper: Nicholls and Gray (2008)
- Missing data and rate heterogeneity: Ryder and Nicholls (2011)
- Lateral trait transfer: Kelly and Nicholls (2017)
- Couplings to diagnose convergence and construct unbiased estimators: Kelly, Ryder and Clarté (2021)
See the TraitLab website for an overview of TraitLab's features and related publications. The manual contains a full description of the tools in TraitLab and a step-by-step tutorial for running an analysis.
TraitLab runs in Matlab and requires its Statistics and Machine Learning toolbox.
The TraitLab source code can be downloaded from GitHub, no further installation steps are required unless the borrowing model is being run: see README in the borrowing
directory for more details.
Please get in touch if you have any issues.
TraitLab reads Nexus-formatted data in a .nex file.
A startup
file sets up the path when Matlab is started at the top level of the TraitLab directory.
To run an experiment using the GUI:
- Start Matlab in the TraitLab folder or start Matlab, navigate to the TraitLab folder and execute
startup
in the Matlab commmand window. - Execute
TraitLab
in the Matlab command window to open the analysis GUI. - Load data and set the parameters of the experiments then click start.
TraitLab will write the settings (.par) and output files to the specified directory.
To analyse samples at the end of a run, open the analysis GUI from the toolbar in the main TraitLab GUI.
Alternatively, experiments can be launched from parameter files.
batchTraitLab('<path to .par file>');
This allows experiments to be run from the Matlab command window or from the command line.
See the manual for a full description.
The example folder contains a synthetic data set and .par files to run single- and coupled-chain experiments. The README file within describes how to run the experiments.