Generalized pruning benchmark repository

This repository is set up so people can reproduce the benchmarking results for the paper A Generalized Phylogenetic Pruning Algorithm by Jun, Nasif, Jennings-Shaffer, Rich, Kooperberg, Fourment, Zhang, Suchard, and Matsen IV.

See the paper for data citations, and please cite the respective papers if you re-use the data.

Prerequisites

• Install bito
  • https://github.com/phylovi/bito
• Install bito GP benchmark environment
  • https://github.com/matsengrp/gp-benchmark-1-environment
• Install Git Large File Storage (LFS) and retrieve DS test data files
  • conda install -c conda-forge git-lfs && git lfs install
  • git lfs pull

Run DS Benchmark

MrBayes MCMC Posterior

• The MrBayes scripts to produce the posterior samples for each dataset are found in ds-benchmark/MrBayesScripts
  • There are two scripts used to sample from the posterior with either uniform or exponential prior on branch lengths.
  • To execute one of the scripts, such as with the uniform prior: bash run-unif-ds.sh
• MrBayes specifications and the posterior tree samples are located in the respective dataset directories found in ds-benchmark/
  • Git LFS is required to retrieve these data files.

Generalized Pruning benchmarking

• Location: ds-benchmark/
  • To run benchmark on all datasets: conda activate bito && bash 0run-all-benchmark-golden.sh
  • To run benchmark on all datasets through a SLURM job: sbatch sbatch-run-all-golden.sh
• Results are output to: ds-benchmark/_golden_benchmark-results/
  • A copy of this results output is saved in the directory manuscript_results/

VBPI Benchmarking

• Location: ds-benchmark/vbpi-exp
• See the following repository for instructions on reproducing these results:
  • https://github.com/matsengrp/vbpi-torch/tree/sample-trees

Run Makona

• The Makona data estimation script is located in makona-benchmark/
  • To run the makona script: conda activate bito && bash run-makona.sh
• Results are output to: makona-benchmark/_output