Leaping through tree space: continuous phylogenetic inference for rooted and unrooted trees

This repo hosts a minimal implementation of GradME

• bme_jax/: Balanced Minimum Evolution and distance-based optimisation with Phylo2Vec in Jax
• cfg/: Example configuration files
• utils/: Utility functions for manipulation of sequence and tree data.

Environment setup

0. Install R (version used here: 4.2.2) if needed. The latest version of R should also work.
1. Setup the gradme environment using conda/mamba and activate the environment:

conda env create -f env.yml
conda activate gradme

2. Optional: if you have GPUs/TPUs, you might need to update your installation of Jax. Follow the instructions at https://github.com/google/jax
3. Install phangorn in R (4.2.2 or above):

install.packages("phangorn")

Accessing data

The following datasets were used:

Dataset	Sites	Taxa	Type	Taxonomic rank	Access	TreeBASE ID
DS1	1,949	27	rRNA (18S)	Tetrapods	[1]	M2017
DS2	2,520	29	rRNA (18S)	Acanthocephalans	[1]	M2131
DS3	1,812	36	mtDNA	Mammals; mainly Lemurs	[1]	M127
DS4	1,137	41	rDNA (18S)	Fungi; mainly Ascomycota	[1]	M487
DS5	378	50	DNA	Lepidoptera	[1]	M2907
DS6	1,133	50	rDNA (28S)	Fungi; mainly Diaporthales	[1]	M220
DS7	1,824	59	mtDNA	Mammals; mainly Lemurs	[1]	M2449
DS8	1,008	64	rDNA (28S)	Fungi; mainly Hypocreales	[1]	M2261
DS9	955	67	DNA	Poaecae (grasses)	[1]	M2389
DS10	1,098	67	DNA	Fungi; mainly Ascomycota	[1]	M2152
DS11	1,082	71	DNA	Lichen	[1]	M2274
Eutherian	1,338,678	37	DNA	Eutherian Mammals	[2]
Jawed	1,460-18,406	99	AA	Gnathostomata (jawed vertebrates)	[3]
Primates	232	14	mtDNA	Mammals; mainly Primates	[4]

• [1] https://bitbucket.org/XavMeyer/coevrj/src/master/data/adaptiveTreeProp/alignments/TreeBase/
• [2] https://datadryad.org/stash/dataset/doi:10.5061/dryad.3629v
• [3] https://datadryad.org/stash/dataset/doi:10.5061/dryad.r2n70
• [4] https://evolution.gs.washington.edu/book/datasets.html
  • Note: We provide the fasta conversion directly here in data/primates.fa

DS1-DS8 are also available at: https://github.com/zcrabbit/vbpi-gnn/tree/main/data/hohna_datasets_fasta DS1-DS11 should also be available on TreeBASE using the TreeBASE, but the site was down on June 9, 2023.

Sources: see manuscript

Running GradME

1. Download the datasets (in the FASTA format) mentioned above and place them in a data/ folder (e.g., in the repo)
2. Update the configuration file cfg/bme_config_v3.yml, especially fasta_path (the path to the FASTA file you want to analyse)
   • (You can also create your own configuration file based on the given template)
3. Run the main optimisation script: python -m bme_jax.main --config-path cfg/name_of_your_config_file.yml or use the demo.ipynb notebook