Installing PICRUSt


Most users will not need to install PICRUSt, but can instead use the online Galaxy version.

PICRUSt is written in python, and has been tested on Mac OS X and Linux systems. The easist way to install PICRUSt is with conda as described below. You will need to download the pre-calculated files before running the tool no matter how it is installed.

Install From Source Step 1. Install Requirements

Follow the install instructions found on the website of each of the dependencies below to install PICRUSt’s dependencies.


h5py needs to be installed manually (numpy is also required, but it should be installed automatically with this package):

  • h5py (version 2.7.1)

You can install h5py with this command:

pip install h5py==2.7.1

Rebuilding PICRUSt’s precalculated 16S rRNA OR Genome Predictions (optional)

  • R installed with APE library

Install From Source Step 2. Download PICRUSt

Release software

The latest release of PICRUSt is picrust-1.1.4.

We recommend the release version of PICRUSt for most users. If you’re not sure whether you want the release or the development version of PICRUSt, you should likely go with the release version.

Development software

Alternatively you can download the latest development version of PICRUSt. You can download the development version of PICRUSt from this link, or using the following command if you have git installed:

git clone git:// picrust

Install From Source Step 3. Install PICRUSt

After downloading PICRUSt, you’ll need to unzip the file. If you’ve downloaded the release version, do this with the following command:

tar -xzf picrust-1.1.4.tar.gz

You’ll then change into the new picrust-1.1.4 directory as follows:

cd picrust-1.1.4

And finally, you’ll install PICRUSt with the following command:

pip install .

These dependencies are automatically installed with pip install .:

Install from Source Step 4. Download PICRUSt’s precalculated files

PICRUSt precomputes most of the computationally intensive pipeline so each user can get predictions with less steps. These files are changed whenever there is a new release of the GreenGenes tree and/or a new release of IMG. These files are fairly large and need to be downloaded separately. Also, you only need the version that corresponds to the GreenGenes that you picked OTUs against (see Picking OTUs for use in PICRUSt).

Assuming you will be picking OTUs against the newest version of GreenGenes and want KEGG Ortholog predictions, then the default files will be sufficient. If you picked against an older version of GreenGenes or are interested in different functional predictions (e.g. COGs) then see the complete list of PICRUSt’s Precalculated Files.

You can install the default files automatically by running:

Alternatively, download these files and place them in the data directory where PICRUSt was installed (e.g. if installed in a conda enviroment named “picrust1”, place the files in $HOME/miniconda3/envs/picrust1/lib/python2.7/site-packages/picrust/data/):