Historical biological data shows that the distribution of phenotypes (biological traits such as height, mass, etc.) along a phylogenetic tree evolves over time, with the rate of evolution not being constant, but rather also including periods of rapid evolutionary ’jumps’. Given observations at the the leaves of a phylogenetic tree, we wish to identify where these jumps have occured.
We follow previous work with a non-parametric Bayesian model based on the Pitman-Yor Process, modelling each node distribution as a Gaussian mixture model with a Pitman-Yor prior. Distributions after a jump on the tree are centered on the previous node, with this hierarchy reflecting the connection between species within the tree.
Using pip, install PhyloJumps to your current environment by running
pip install git+https://github.com/shimlab/PhyloJumps.gitThis is equivalent to cloning the environment and building the package.
git clone https://github.com/shimlab/PhyloJumps.git
cd PhyloJumps
python -m pip install .Included in the current repository is also a conda environment.yaml file to replicate the author's Python 3.8 environment.
conda env create --file=environment.yamlThe default name of the conda environment is phylojumps.
See examples.
TODO