Installation¶
Setting up a virtual environment first¶
We highly recommended installing marsi inside a virtual environment using (anaconda).
There is no conda recipe for marsi, but the dependencies can be easily installed using conda.
MARSI needs to be installed it using pip. To create a virtual environment and get
some of the heavier dependencies out of the way, run the following command:
$ conda create -y -n marsi python=3.4 cython
Alternatively you can use virtualenv. virtualenvwrapper tremendously simplifies using virtualenv and can easily
be installed using virtualenv-burrito. Once you installed virtualenv and virtualenvwrapper, run
$ mkvirtualenv marsi # or whatever you'd like to call your virtual environment
$ workon marsi
and then continue with the installation instructions described below.
Dependencies required before installation¶
Using conda¶
All dependencies will by installed by running the following commands
$ conda install -c openbabel openbabel
$ conda install -c rdkit rdkit
Alternatively you can install the dependencies one by one.
Cython¶
To install Cython follow the cython-installation documentation provided on their website.
Numpy¶
Numpy can be installed using conda:
$ conda install numpy
Or can also be installed using pip:
$ pip install numpy
Eigen3¶
To install Eigen3 on Ubuntu/Debian run:
$ sudo apt-get install libeigen3-dev
To install Eigen3 on MacOS X use homebrew:
$ brew install eigen3
OpenBabel¶
To install OpenBabel on Ubuntu/Debian run:
$ sudo apt-get install openbabel libopenbabel-dev
On MacOS X it can be installed using homebrew:
$ brew install open-babel
RDKit¶
RDKit can be installed using the rdkit-documentation.
For MacOS X we recommend using homebrew:
$ brew install rdkit
GLPK¶
Using constraint-based methods with marsi requires glpk to be installed. In order to generate python bindings, swig is also required. On Ubuntu/Debing we recommend using:
$ sudo apt-get install libglpk-dev glpk-utils swig
On MacOS X it can be installed using homebrew.
$ brew install swig
$ brew install glpk
Optional¶
CPLEX¶
CPLEX is a commercial solver with great performance. It solves linear, mixed-integer and quadratic problems. Install CPLEX as described in cplex-install and then install the python bindings (cplex-python).
Soft dependencies¶
The following soft dependencies can be installed all at once using pip install marsi[all] or individually
by specifying individual categories of dependencies (for example pip install marsi[jupyter,3d, ...]).
The following categories are available:
'docs': ['Sphinx>=1.3.5', 'numpydoc>=0.5'],
'jupyter': ['jupyter>=1.0.0', 'ipywidgets>=4.1.1'],
'test': ['nose>=1.3.7', 'rednose>=0.4.3', 'coverage>=4.0.3'],
'3d': ['imolecule>=0.1.13'],
'opencl': ['pyopencl>=2016.1']