These are installation instructions for a Debian/Ubuntu Linux
system and Mac OS X. A Windows installation can also be done by
following the Mac instructions if a suitable C++ compilation
environment is available.

See also https://galen.dulci.duhs.duke.edu/flow/wiki/Software

Downloading Flow software
=========================

Download by clicking on the bz2, zip or gz links at the top of http://hg/dulci.org/flow

Debian/Ubuntu installation
==========================

The simplest way to install the package is to download the flow-x.x.x-i386.deb or flow-x.x.x-amd64.deb from https://galen.dulci.duhs.duke.edu/flow/wiki/Software, and install using a package manager, for example

sudo gdebi flow-x.x.x-i386.deb

This will automatically install most of the dependencies needed. However the python-mdp package is only available for Debian lenny and sid, and has not made it to the Ubuntu repositories yet, so manual installation is still necessary by following these steps

Download tar.gz file from sourceforge http://sourceforge.net/project/showfiles.php?group_id=116959
tar -xzf MDP-2.3.tar.gz 
sudo python setup.py install

[[ Base system ]] 
sudo apt-get install python-wxgtk2.6 python-numpy python-tables python-scipy python-matplotlib python-networkx

[[ Plugins ]]

[ Graphics/ThreeDFrame ]
sudo apt-get install python-vtk

[ Statistics/kmeans ]
sudo apt-get install python-biopython

[ Statistics/Kde and Statistics/bayes ]
sudo apt-get install scons libboost-dev r-mathlib blitz++ lapack3-dev

# install the ANN library (for kde)
cd /tmp
wget http://www.cs.umd.edu/~mount/ANN/Files/1.1.1/ann_1.1.1.tar.gz
tar -xzf ann_1.1.1.tar.gz
cd ann_1.1.1/
make linux-g++
cd ..
sudo mv ann_1.1.1 /usr/local/src

# install the boost numeric binding headers (for bayes)
cd /tmp
wget http://news.tiker.net/news.tiker.net/download/software/boost-bindings/boost-bindings-20070826.tar.gz
tar -xzf boost-bindings-20070826.tar.gz
sudo cp -R boost-bindings/boost/numeric/bindings/ /usr/include/boost/numeric/

# alternatively, check out svn repository of boost-sandbox
svn co http://svn.boost.org/svn/boost/sandbox/boost/numeric/bindings ublas_extensions/boost/numeric/bindings
sudo cp -R ublas_extensions/boost/numeric/bindings/ /usr/include/boost/numeric/

# build the python shared libraries for kde and Bayes where $FLOW is the base directory
sudo apt-get install libboost-python-dev
cd $FLOW/src
scons install

[ Projections/ica ]
sudo apt-get install python-rpy r-base gfortran
sudo R
> install.packages("fastICA")
> q()

# python-rpy binaries are tied to specific R versions
# if the version from the package manager fails (e.g. on Ubuntu 8.02)
# uninstall and install from source
sudo apt-get remove python-rpy
cd /tmp
wget superb-west.dl.sourceforge.net/sourceforge/rpy/rpy-1.02.tar.gz
tar -xzf rpy-1.0.2.tar.gz
cd rpy-1.0.2
python setup.py build
python setup.py install

[Projections/ica2 and pca2 using the Python Modular toolkit]
# available for Debian lenny and sid
sudo apt-get install python-mdp

# all other distributions
Download tar.gz file from sourceforge http://sourceforge.net/project/showfiles.php?group_id=116959
tar -xzf MDP-2.3.tar.gz 
sudo python setup.py install

# Flow should now be ready to run
cd $FLOW/src
python App.py

Fedora Core 9 installation
==========================

Packages available using yum
-----------------------------
sudo yum install numpy scipy matplotlib hdf5-devel wxPython vtk-python python-setuptools R R-devel rpy 

Packages to install from source
-------------------------------

1. pytables
mkdir src
cd src
wget http://www.pytables.org/download/stable/pytables-2.0.4.tar.gz
tar xzf pytables-2.0.4.tar.gz
cd pytables-2.0.4
python setup.py build_ext --inplace
sudo python setup.py install
exit
cd ..

2. networkx
wget https://networkx.lanl.gov/download/network/networkx-0.37.tar.gz
tar -xzf networkx-0.37.tar.gz
cd networkx-0.37
python seutp.py build
sudo python setup.py install
exit
cd ..

3. MDP
wget http://voxel.dl.sourceforge.net/sourceforge/mdp-toolkit/MDP-2.3.tar.gz
tar xvf MDP-2.3.tar.gz
cd MDP-2.3
python setup.py build
sudo python setup.py install
exit
cd ..

4. PyCluster
wget http://bonsai.ims.u-tokyo.ac.jp/~mdehoon/software/cluster/Pycluster-1.41.tar.gz
tar xzf Pycluster-1.41.tar.tz
cd Pycluster-1.41

5. fastICA using R (optional but useful as an example of R integration)
sudo yum install gcc-gfortran
R
> install.packages('fastICA')
Follow prompts and quit when done

Download and run Flow
---------------------

wget http://hg.dulci.org/flow/archive/tip.tar.gz
tar xzf tip.tar.gz
cd flow-xxxxxxxxxxxxxx (where xxxxxxxxxxxxxx is some machine-assigned label)
cd src
python App.py


OpenSUSE 10.3 64-bit installation
=================================

Add the following repositories to YAST (or other package management tool)
------------------------------------------------------------------------

http://download.opensuse.org/repositories/science/openSUSE_10.3/
http://software.opensuse.org/download/repositories/science/openSUSE_10.3/

Install the following using YAST (or other package management tool)
------------------------------------------------------------------

gcc gcc-c++ make python-numpy python-scipy python-setuptools
python-tables python-wxgtk python-matplotlib python-matplotlib-wx
R-base 

Install the following python packages using setuptools
------------------------------------------------------

sudo easy_install networkx

Install the following from source
---------------------------------

# python Cluster library
 wget http://bonsai.ims.u-tokyo.ac.jp/~mdehoon/software/cluster/Pycluster-1.41.tar.gz
tar -xzf Pycluster-1.41.tar.gz 
cd Pycluster-1.41/
sudo python setup.py install

# R fastICA library
R

# ANN library (for kde)
cd /tmp
wget http://www.cs.umd.edu/~mount/ANN/Files/1.1.1/ann_1.1.1.tar.gz
tar -xzf ann_1.1.1.tar.gz
cd ann_1.1.1/
make linux-g++
cd ..
sudo mv ann_1.1.1 /usr/local/src

# python MDP library
wget http://voxel.dl.sourceforge.net/sourceforge/mdp-toolkit/MDP-2.3.tar.gz
tar -xzf MDP-2.3.tar.gz 
sudo python setup.py install

# VTK (python bindings do not appear to work in 64-bit openSUSE 10.3)
wget http://www.vtk.org/files/release/5.2/vtkdata-5.2.0.tar.gz
tar -xzf vtkdata-5.2.0.tar.gz 
wget http://www.vtk.org/files/release/5.2/vtk-5.2.0.tar.gz
tar -xzf vtk-5.2.0.tar.gz
mkdir VTKBuild
cd VTKBuild
ccmake ../VTK (set the options as shown)
 BUILD_EXAMPLES                   ON                                           
 BUILD_SHARED_LIBS                ON                                           
 BUILD_TESTING                    OFF                                          
 CMAKE_BACKWARDS_COMPATIBILITY    2.4                                          
 CMAKE_BUILD_TYPE                                                              
 CMAKE_INSTALL_PREFIX             /usr/local                                   
 VTK_DATA_ROOT                    < VTKData location - auto-detected >
 VTK_USE_INFOVIS                  ON                                           
 VTK_USE_PARALLEL                 OFF                                          
 VTK_USE_RENDERING                ON                                           
 VTK_USE_VIEWS                    ON                                           
 VTK_WRAP_JAVA                    OFF                                          
 VTK_WRAP_PYTHON                  ON                                           
 VTK_WRAP_TCL                     OFF
make
sudo make install




Mac OS X 10.5  installation
=====================

[[ Base system ]]

Download and install the wxPython binary from http://www.wxpython.org/

Install the binaries for scipy, numpy and matplotlib from
http://pythonmac.org/packages/py25-fat/index.html

cd /tmp
wget http://pypi.python.org/packages/source/n/networkx/networkx-0.36.tar.gz
tar -xzf networkx-0.36.tar.gz 
cd networkx-0.36
python setup.py build
sudo python setup.py install

cd /tmp
wget http://www.pytables.org/download/stable/pytables-2.0.3.tar.gz
tar -xzf pytables-2.0.3.tar.gz 
cd pytables-2.0.3
python setup.py build
sudo python setup.py install


[[ Plugins ]]

[ Graphics/ThreeDFrame ]

* Installing cmake
cd /tmp
wget http://www.cmake.org/files/v2.4/cmake-2.4.8.tar.gz
tar -xzf cmake-2.4.8.tar.gz
cd cmake-2.4.8
./bootstrap
make
sudo make install

* Installing VTK
cd /tmp
cvs -d :pserver:anonymous@public.kitware.com:/cvsroot/VTK login
(respond with password vtk)
cvs -d :pserver:anonymous@public.kitware.com:/cvsroot/VTK checkout VTK
mkdir VTKBuild
cd VTKBuild
ccmake ../VTK
options (BUILD_SHARED_LIBS ON; VTK_WRAP_PYTHON ON)
Hit 'c' to configure and 'g' to generate Makefiles
make
sudo make install
cd Wrapping/Python
python setup.py build
sudo python setup.py install
export PYTHONPATH=/Library/Python/2.5/site-packages/VTK-5.1.0-py2.5.egg/vtk

[ Statistics/kmeans ]
wget http://downloads.sourceforge.net/numpy/Numeric-24.2.tar.gz

[ Statistics/Kde and Statistics/bayes ]

# install the ANN library (for kde)
cd /tmp
wget http://www.cs.umd.edu/~mount/ANN/Files/1.1.1/ann_1.1.1.tar.gz
tar -xzf ann_1.1.1.tar.gz
cd ann_1.1.1/
make macosx-g++
cd ..
sudo mv ann_1.1.1 /usr/local/src

# install the blitz++ library
cd /tmp
wget http://internap.dl.sourceforge.net/sourceforge/blitz/blitz-0.9.tar.gz
tar -xzf blitz-0.9.tar.gz
cd blitz-0.9
./configure
make
sudo make install

# install the C++ boost library
cd /tmp
wget http://superb-west.dl.sourceforge.net/sourceforge/boost/boost_1_35_0.tar.gz
tar -xzf boost_1_35_0.tar.gz
cd boost_1_35_0
./configure
make
sudo make install

# install the boost numeric binding headers (for bayes)
cd /tmp
wget http://mathema.tician.de/news.tiker.net/download/software/boost-numeric-bindings/boost-numeric-bindings-20081116.tar.gz
tar xzf boost-numeric-bindings-20081116.tar.gz 
sudo cp -R boost-bindings/boost/numeric/bindings/ /usr/include/boost/numeric/

# alternatively, check out svn repository of boost-sandbox
cd /tmp
svn co http://svn.boost.org/svn/boost/sandbox/boost/numeric/bindings ublas_extensions/boost/numeric/bindings
sudo cp -R ublas_extensions/boost/numeric/bindings/ /usr/include/boost/numeric/

# build the python shared libraries for kde and Bayes where $FLOW is the base directory
cd $FLOW/src
scons install

[ Projections/ica ]

# Install R from source
cd /tmp
wget http://cran.r-project.org/src/base/R-2/R-2.6.2.tar.gz
tar -xzf R-2.6.2.tar.gz
cd R-2.6.2.tar.gz
./configure --enable-BLAS-shlib --with-blas='-framework vecLib'
make
sudo make install
cd src/nmath/standalone
make
sudo ln -s /Library/Frameworks/R.framework/Resources/R /usr/local/bin/R

# Install fastICA
sudo R
> install.packages("fastICA")

# Install rpy
cd /tmp
wget superb-west.dl.sourceforge.net/sourceforge/rpy/rpy-1.02.tar.gz
tar -xzf rpy-1.0.2.tar.gz
cd rpy-1.0.2
python setup.py build
python setup.py install

[Projections/ica2 and pca2 using the Python Modular toolkit]
Download tar.gz file from sourceforge http://sourceforge.net/project/showfiles.php?group_id=116959
tar -xzf MDP-2.3.tar.gz 
sudo python setup.py install

Windows Installation
====================

Note - this was done on a Windows Vista Business Machine with User
Access Control (UAC) turned off.

The following binaries have nice Windows installers

Download and install the Enthought binary from http://www.enthought.com/products/epddownload.php
Download and install Numeric binary from http://biopython.org/DIST/Numeric-24.2.win32-py2.5.exe
Download and install PyCluster binary from http://bonsai.ims.u-tokyo.ac.jp/~mdehoon/software/cluster/Pycluster-1.39.win32-py2.5.exe
Download and install Networkx binary from http://pypi.python.org/packages/any/n/networkx/networkx-0.36.win32.exe

Add Python to the Environment path by clicking on the Start icon, then
right-clicking on Computer and then "Advanced system settings" on the
Tasks pane which will open a System Properties dialog. Click "Environment Variables" and edit "System variables" and add

 ;C:\Python25;C:\Python25\Scripts

to the end of the Path string.

== Optional install of MDP for ICA and PCA functionality ==
MDP also has a Windows binary at
http://downloads.sourceforge.net/mdp-toolkit, but the link was down
when I tried it. If it works for you, install the binary.

If not , download MDP from
http://downloads.sourceforge.net/mdp-toolkit/MDP-2.3.zip 

Unzip into an appropriate directory, open a Command Prompt, navigate
to the MDP-23 folder and type "python setup.py install"

== Note ==
Executables for Windows 32 and 64 bit DLLs need to be provided, and the Scons script modified to include a section for Windows installations.