Installation

The PICRUSt2 beta version is available as a standalone version or as a QIIME2 plugin. Both versions will work on Linux and Mac OS X. To use this program on Windows you would first need to install a Linux emulator.

Note that there are fewer options currently available for the QIIME2 plugin, but it is easier to run on personal computers (in particular there are lower memory requirements and it will run on Mac laptops). A tutorial on this QIIME2 plugin is available here and the alternative installation instructions for installation and use of PICRUSt2 with this plugin are here.

Please note that generating stratified output tables with the PICRUSt2 pipeline (predictions broken down by contributing sequences) requires more computing resources than the PICRUSt1 metagenome prediction steps! You will need at least 16 GB of RAM to generate stratified output tables and the running time may be several hours depending on how many CPUs you can dedicate (and how many samples and sequences are input).

You can install PICRUSt2 either with conda or from source, which are both shown below. Note that both approaches require that you first install conda.

Install from bioconda

Create a new environment with PICRUSt2 installed and activate the environment. If you receive an error about the default reference files missing when you try to run the program you will need to install from source instead.

conda create -n picrust2 -c bioconda -c conda-forge picrust2=2.1.4_b
conda activate picrust2

Install from source

Download the source tarball, untar, and move into directory. You can check that you are downloading the latest release here: https://github.com/picrust/picrust2/releases

wget https://github.com/picrust/picrust2/archive/v2.1.4-b.tar.gz
tar xvzf  v2.1.4-b.tar.gz
cd picrust2-2.1.4-b/

Create and activate the environment (with requirements) and then install PICRUSt2 with pip.

conda env create -f  picrust2-env.yaml
conda activate picrust2
pip install --editable .

Run the tests to verify the installation (note that this wont work if you followed the bioconda install instructions):

pytest