Before starting, in order to use pip, users will want to load a later version of Python:

module load python

Users may want to install python packages that are frequently used by their groups, in a common location to the user or group.

The simple way installs under .local in your home directory (substitute package-name for what you want):

pip install --user package-name

or, if you download the package and it has a

python --user

If you’d like to install the package to a custom directory, here is a more detailed example:

  1. Create a local directory:
    mkdir -p $HOME/my-python-modules/lib/python2.7/site-packages
  2. Download the package source, preferably to the scratch directory. This example will install “pyquante”. This package can be downloaded from “”. You may be able to use wget or svn, etc, to download your package.
  3. cd $HOME/scratch
  4. tar zxvpf PyQuante-1.6.5.tar.gz

    This will extract all the files into /scratch/users/$USER/PyQuante-1.6.5

  5. cd PyQuante-1.6.5
  6. Type “ls”. You should see a file “”
  7. Make sure you load the version of python you want to use, for example: module load python
  8. Check: type “ml” You should see Python/2.7.10
  9. Install it on your local directory:
    python install --prefix=$HOME/my-python-modules 

    Note: You may want to check that you are using the proper compiler.

  10. Check:
    ls $HOME/my-python-modules/lib/python2.7/site-packages
  11. Utilization. You need to either do this manually or create a module
  12. Manually:
    export PATH=$HOME/my-python-modules/bin:$PATH (if necessary)
     export PYTHONPATH=$HOME/my-pyhton-modules/lib/python2.7/site-packages:$PYTHONPATH
  13. Create a module.
    "mkdir -p $HOME/privatemodules/PyQuante"
  14. Create a file called “1.6.5.lua” in the above directory with the following contents:  (NOTE:  1.6.5 is the version number in this case for PyQuante)
    whatis([[local module for PyQuante ]])
    append_path{"PYTHONPATH", "/$HOME/my-python-modules/lib/python2.7/site-packages", delim=":",priority="0"}
  15. To load it: “module load use.own”
    module load PyQuante
  16. Check. echo $PYTHONPATH



That is it!