Many tools don’t require root privileges and can be installed in user space. This enables users to customize Python packages to their needs, without conflicting with any system-level Python installations, or Python installations other users have performed.
A Miniconda Python installation is a common prerequisite for Python-based applications in machine learning and other fields.
Miniconda is a free software package that includes the condapackage and environment manager. https://docs.conda.io/en/latest/miniconda.html
1. Download the Miniconda installer
2. Run the installer
$chmod 755 Miniconda3-latest-Linux-x86_64.sh
3. Configuring Conda
$./conda update conda
$./condaconfig --add channels bioconda
Miniconda installers contain the conda
package manager and Python
$ condainstall numpy
Creating environments with Conda
$ condainstall bwa
Or a new environment can be created:
$ condacreate -n aligners bwa bowtie hisatstar
Many HPC groups are using containers for their software. Some tools come as singularity or docker images.
(1) If there is no docker container available for your package you can build one in your local machine with the OS of your choice and install your package.
(2) Push the container to a cloud container repo. ie. https://hub.docker.com
(3) Pull the docker container as a singularity image in juno, doing
singularity pull docker://<your-image>
(4) Run your package by doing
singularity run <your-image> <your-package-command>
Module avail will show you the available packages and versions.
You can use module to load singularity as:
module load singularity/3.7.1
Please read more about how to use containers.
If you would like to request help installing the software, please use this form here: http://hpc.mskcc.org/software-installation-request/