The easiest way to install BioSimSpace is using our conda channel:
conda install -c rdkit -c conda-forge -c omnia -c michellab biosimspace
To install the latest development version you can use:
conda install -c rdkit -c conda-forge -c omnia -c michellab/label/dev biosimspace
If you plan on using BioSimSpace interactively via Jupyter, then you might also need to enable the required notebook extensions within your Conda environment:
jupyter-nbextension enable fileupload --py --sys-prefix jupyter-nbextension enable nglview --py --sys-prefix
Unless you add the required channels to your Conda configuration, then you’ll need to add them when updating, e.g., for the development package:
conda update -c rdkit -c conda-forge -c omnia -c michellab/label/dev biosimspace
Note that because of Conda’s peculiar scoring metrics you might not end up with the latest version of BioSimSpace when performing a fresh install or update. (It tries to minimise various things, such as the number of dependencies installed, which is difficult when your package depends on many other packages.) To see what packages are available, run:
conda search -c michellab/label/dev biosimspace
You can then install the latest version by explicitly stating the full package name, e.g.:
conda install -c rdkit -c conda-forge -c omnia -c michellab/label/dev biosimspace=2019.1.0=py37h14c3975_85
On OS X you will need to run Python scripts with the
interpreter. This is due to an issue with the default Python interpreter that
is installed via Conda. (This applies to all installation methods.)
The self-extracting binary for the 2019.1.0 release of BioSimSpace can be downloaded from one of the following links:
The latest self-extracting binary for the development version of BioSimSpace can be downloaded from one of the following links:
(These are portable X86-64 binaries that should work on any Linux distribution released since ~2011, or any OS X >= 10.9 [Mavericks, released 2013]. Note that they are compiled with AVX enabled, so will only work on modern (>2011) X86-64 Intel/AMD processors.)
Once downloaded, the binary can be unpacked as follows, e.g. for the Linux development package:
chmod +x biosimspace_devel_latest_linux.run ./biosimspace_devel_latest_linux.run
This will let you choose where to install BioSimSpace. By default, this will be
The following documents a full installation of BioSimSpace from source. Before starting, you’ll need a working Git installation.
BioSimSpace is built on top of the Sire molecular simulation framework. To download and install Sire:
git clone https://github.com/michellab/Sire cd Sire ./compile_sire.sh
Assuming the default installation path, this will install Sire into
(Note that the installation is slow and can take in excess of an hour.)
Next you will need to download BioSimSpace and install it into your Sire application. (The following assumes the default Sire installation path.)
git clone https://github.com/michellab/BioSimSpace cd BioSimSpace/python $HOME/sire.app/bin/python setup.py install
Once finished, you can test the installation by running:
Then try importing the BioSimSpace package:
import BioSimSpace as BSS
When developing you may not wish to continually re-install BioSimSpace and its
associated dependencies. To avoid this, you can either make use of
PYTHONPATH=$HOME/Code/BioSimSpace/python $HOME/sire.app/bin/python script.py
or use the
develop argument when running the
setup.py script, i.e.
PYTHONPATH=$HOME/sire.app/bin/python setup.py develop
You can also skip installation of external dependencies by setting the
BSS_SKIP_DEPENDENCIES=True $HOME/sire.app/bin/python setup.py install
Some BioSimSpace functionality requires OpenMM. Although a bundled version is provided as part of the installation, this may not be appropriate for your GPU drivers. To automatically detect and install a suitable version of OpenMM, simply run the following command post-install:
(Note that, depending on your installation method,
be located in
Alternatively, to manually install a particular version of OpenMM you can use a specific Conda label, e.g.:
conda install -c omnia/label/cuda90 openmm
If you have compiled Sire against a custom OpenMM installation, then you’ll
need to set the
OPENMM_PLUGIN_DIR environment variable to point to the
correct plugin location. By default this variable is set to the plugin
directory of the bundled OpenMM package.
If you experience problems with Matplotlib when importing BioSimSpace on macOS, e.g.
RuntimeError**: Python is not installed as a framework.
simply add the following to
Note that plotting functionality will be disabled if you are using BioSimSpace on a remote server without X forwarding.
If you experience problems with Jupyter permissions, try removing
Several additional packages are required for full access to all of BioSimSpace’s functionality. Please download and install these packages according to their recommended installation instructions.