Installation

Requirements

The following dependencies are needed to run the code (as-written). Some details on installing them are given below:

Package Source PyPI
numpy Source PyPI
scipy Source PyPI
scikits.bootstrap Source PyPI
statsmodels Source PyPI
tqdm Source PyPI
pandas Source PyPI
mayavi Source PyPI
vtk Source  

A Note on VTK

VTK (the Visualization ToolKit) is a system that allows for easy visualization of 3D data. It is only required if you wish to do visualization, and is a dependency of the Mayavi package. Some of the code in the tpb module uses this toolkit to enable a visualization of TPB networks and how the code randomly subsamples the volume.

Windows

On Windows, the easiest way to get going is to use Cristoph Gohlke’s prebuilt binaries. Download the correct binary for your Python version (2.7, 3.4, or 3.5) and Python build (32 or 64 bit). Once downloaded, go to the folder where the file is located, and install using pip. For example:

$ pip install VTK-7.0.0-cp35-cp35m-win_amd64.whl

Linux

On Linux, binary wheel distributions are not provided for packages such as this. You can search for a VTK distribution in your package manager, but chances are it is compiled with Python 2 bindings. This may be fine if you’re using Python 2.7, but you should really be using a Python 3.X version. For Python 3, you will probably have to download the source and compile the package using CMake.

There are some helpful instructions here. The important CMake flags to set to make sure you compile Python wrappers for 3.X are: VTK_WRAP_PYTHON, VTK_PYTHON_VERSION, and then all the PYTHON* flags. Make sure that the wrap Python flag is set, the Python version is set to the version you have (e.g. 3.5), and then click Configure. CMake should find your installed Python version and correctly fill all the PYTHON* flags correctly.

The other flag you want to make sure is set correctly is CMAKE_INSTALL_PREFIX. This needs to be set so that the compiled Python libraries get installed to a place on your Python PATH. On my system, CMake defaulted to setting this to /usr/local, but I needed to change it to /usr, because my Python installation is in /usr/lib/python3.5. Once all the flags are set, click Configure until no more new flags appear, and then click Generate. This will generate the Makefile for the installation.

Change to the build directory and then run the following code (replacing the number 4 with however many cores you have on your machine):

$ make -j4
$ sudo make install

Check that Python can find the installation correctly by running:

$ python -c "import vtk"

If this test does not report any errors, then your installation probably went okay.

Development Version Installation

Currently, the only supported method of installation is to clone the development version and install using pip. The latest version of the code should be available in the online repository. To get this version installed on your system: install the requirements from above, clone the repository, and then install with pip:

$ git clone https://jat255@bitbucket.org/jat255/fibbootstrap.git
$ cd fibbootstrap
$ pip install -e ./

If you would like to use the Avizo scripts as well, those are available as a submodule of this package, and can be obtained by doing a recursive git clone:

$ git clone --recursive https://jat255@bitbucket.org/jat255/fibbootstrap.git