A-pro

ASTRAL-Pro stands for ASTRAL for PaRalogs and Orthologs. ASTRAL is a tool for estimating an unrooted species tree given a set of unrooted gene trees and is statistically consistent under the multi-species coalescent model (and thus is useful for handling incomplete lineage sorting, i.e., ILS). ASTRAL-pro extends ASTRAL to allow multi-copy genes. ASTRAL-pro finds the species tree that has the maximum number of shared induced quartet tree equivalent classes with the set of gene trees, subject to the constraint that the set of bipartitions in the species tree comes from a predefined set of bipartitions. Please see the paper below for the definition of the PL-quartet scores, which is what ASTRAL-Pro optimizes. We refer to the tool both as A-Pro and ASTRAL-Pro.

Publications:

Chao Zhang, Siavash Mirarab, ASTRAL-Pro 2: ultrafast species tree reconstruction from multi-copy gene family trees, Bioinformatics, 2022, btac620, https://doi.org/10.1093/bioinformatics/btac620

Chao Zhang, Celine Scornavacca, Erin K Molloy, Siavash Mirarab, ASTRAL-Pro: Quartet-Based Species-Tree Inference despite Paralogy, Molecular Biology and Evolution, Volume 37, Issue 11, November 2020, Pages 3292–3307, https://doi.org/10.1093/molbev/msaa139

*** IMPORTANT NOTICE: PLEASE READ ***

Please switch to ASTRAL-Pro2, as the current version of ASTRAL-Pro will no longer be maintained! ASTRAL-Pro2, an equally accurate yet faster and easier to install version of A-pro with lower memory consumption is implemented in c++ here and published here.

Required installations

• Java (>=1.7)

• If you need to compile (you may not), you will also need JDK and g++ or icc (supporting c++11 standard)

Installation

Download:

You simply need to download the zip file (under clone and download zip) and extract the contents to a folder of your choice. Alternatively, you can clone the github repository.

Normally, you would not need to install. You can just run A-Pro, as we show bellow.

If you have trouble with running the packaged A-Pro,

1. cd ASTRAL-MP
2. bash make.sh.

For more information please see AVX2 section of https://github.com/smirarab/ASTRAL/tree/MP

If it is hard to install or run ASTRAL-Pro on your machine, you can try an unpublished C++ version here.

Input requirement

The input gene trees must be in the Newick format, either multi-labelled (eg. ((SpeciesA,SpeciesB),(SpeciesA,SpeciesC));) or with gene-name-to-species-name mapping provided in the following format (eg. for ((GeneA1,GeneB1),(GeneA2,GeneC1));):

GeneA1 SpeciesA
GeneA2 SpeciesA
GeneB1 SpeciesB
GeneC1 SpeciesC

Taxon names cannot have quotation marks in their names (sorry!). This means you also cannot have weird characters like ? or | in the name (underscore is fine).　A-pro is not designed for polytomies and will randomly resolve them during pre-processing.

Output options

The output in is Newick format and gives:

• the species tree topology,
• branch lengths in coalescent units,
• branch supports measured as local posterior probabilities.
• It can also annotate branches with other quantities, such as quartet support, as described in the tutorial.

The ASTRAL tree leaves the branch length of terminal branches empty. Some tools for visualization and tree editing do not like this (e.g., ape). In FigTree, if you open the tree several times, it eventually opens up (at least on our machines). In ape, if you ask it to ignore branch lengths all together, it works. In general, if you tool does not like the lack of terminal branches, you can add a dummy branch length, as in this script.

Running A-pro

cd to ASTRAL-MP and run for multi-labelled gene trees:

java -D"java.library.path=lib" -jar astral.<version_number>.jar -i input

The results will be outputted to the standard output. To save the results in a file use the -o option (Strongly recommended):

java -D"java.library.path=lib" -jar astral.<version_number>.jar -i input -o output

To save the logs (also recommended), run:

java -D"java.library.path=lib" -jar astral.<version_number>.jar -i input -o output 2>out.log

cd to ASTRAL-MP and run for gene trees with gene-name-to-species-name mapping provided:

java -D"java.library.path=lib" -jar astral.<version_number>.jar -i input -a mapping -o output

Note that instead of cd to ASTRAL-MP, you can replace astral.<version_number>.jar and -D"java.library.path=lib/" with absolute path to A-Pro. For example, if my A-Pro is located on /Users/chaoszhang/A-pro, you can use:

java -D"java.library.path=/Users/chaoszhang/A-pro/lib" -jar /Users/chaoszhang/A-pro/astral.<version_number>.jar -i input -o output

Examples:

cd to ASTRAL-MP and run for multi-labelled gene trees:

java -jar -D"java.library.path=lib" astral.1.1.2.jar -i ../example/example1.tre

cd to ASTRAL-MP and run for gene trees with gene-name-to-species-name mapping provided:

java -jar -D"java.library.path=lib" astral.1.1.2.jar -i ../example/example2.tre -a ../example/example2map.txt

Handling very large datasets

Memory:

For big datasets (say more than 500 taxa), increasing the memory available to Java can result in speedups. Note that you should give Java only as much free memory as you have available on your machine. So, for example, if you have 3GB of free memory, you can invoke ASTRAL using the following command to make all the 3GB available to Java:

java -Xmx3000M -D"java.library.path=lib/" -jar astral.<version_number>.jar -i input

Search Space:

For big datasets (say more than 500 taxa), generating search space may take more than reasonable time. Reducing excessive search space can result in speedups. The search space can be controlled by placing astral-pro.config file in your working directory (the directory you get when you run pwd) and then changing values of X, Y, and Z. The file has the following format:

X 
Y 
Z

Each multi-copy gene tree will be broken into 2^Z single-copy trees; only single-copy trees with more than X leaves will be included in the search space and only single-copy trees with more than Y leaves will be used to generate extra search space. You reduce the search space, you want to increase X and Y to be close to one third or half of the number of species you have. For example, for a dataset with 1200 species and 5000 genes, we used the following values with success.

200
100
2

Acknowledgment

ASTRAL code reuses code from ASTRAL-MP and bytecode and some reverse engineered code from PhyloNet package (with permission from the authors).

Bug Reports:

contact astral-users@googlegroups.com