INSTALLATION.txt

========================================================================================

UPDATE: ATTRACT uses Python 2, which is reaching its end of life
The easiest way to use ATTRACT is from a Docker container
Install Docker, and then obtain the ATTRACT image :
    docker pull rpbs/attract
Then run your ATTRACT shell script (from the web interface, or hand-written): 
    ./run-in-docker.sh my-attract-shell-script.sh

========================================================================================


ATTRACT is developed under Linux. It should work under Mac OSX, but this has not been tested.
It does not work under Microsoft Windows.
Note that ATTRACT's Python scripts are written for Python2.7, not for Python3.

Installation instructions for Linux:

1. Install g++ and gfortran (under Ubuntu: sudo apt-get install g++ gfortran).

2. Go into "attract/bin", "make all" or "make all -j4" (if you have 4 cores).
This should create the attract binary. See step 7 to compile ATTRACT with support for GPU acceleration.

3. Export the ATTRACT environment variables. In your .bashrc file, add the following:
  export ATTRACTDIR=/home/user/attract/bin (i.e. wherever you installed attract)
  export ATTRACTTOOLS=$ATTRACTDIR/../tools
  export ATTRACTGUI=$ATTRACTDIR/../gui
  export LD_LIBRARY_PATH=$ATTRACTDIR:$LD_LIBRARY_PATH

4. For the analysis, also install numpy and scipy (under Ubuntu: sudo apt-get install python-numpy python-scipy)

5. For iATTRACT and the OPLSX forcefield, install pdb2pqr (under Ubuntu: sudo apt-get install pdb2pqr)
In addition, the pdb2pqr directory must be made accessible to Python. (Under Ubuntu, this directory is
/usr/share/pdb2pqr . It should contain the files "pdb2pqr.py" and "main.py".)
To make it accessible, edit your .bashrc file (/home/user/.bashrc) by adding the following
  export PYTHONPATH=$PYTHONPATH:/usr/share/pdb2pqr

6. Test your installation.
Type "source ~/.bashrc" (or start a new command line shell).
After step 1 and 2, typing "$ATTRACTDIR/attract" should result in:
  Too few arguments
  usage: $path/attract structures.dat parameterfile receptor.pdb [ligand.pdb] [options]
After step 3, typing "python -c 'import numpy, scipy' " should give no error.
After step 4, typing "python -c 'import pdb2pqr' " should give no error.

Optional steps:

7. GPU support
- Install the CUDA toolkit. For more information visit "https://developer.nvidia.com/cuda-toolkit".
On Ubuntu (17.10 or later), you can do the following:
sudo apt-get install nvidia-cuda-toolkit
and then add the following to your .bashrc:
    export CUDADIR=/usr/lib/nvidia-cuda-toolkit
    export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$CUDADIR/lib64
- Install the boost coroutine and system libraries (under Ubuntu: sudo apt-get install libboost-coroutine-dev libboost-system-dev libboost-program-options-dev).
- Repeat step 2. This should generate the emATTRACT, mcATTRACT and gpuATTRACTcheck binaries.
- Check that GPU support is enabled successfully by running the gpuATTRACTcheck binary.

9. PyPy
Some ATTRACT tools will run considerably faster if you install PyPy
(under Ubuntu: sudo apt-get install pypy)

10. Spydersilk support
With Spydersilk, you can edit ATTRACT protocol files (.web files) locally.
- Get the Spydersilk code (under Ubuntu: wget http://www.attract.ph.tum.de/services/ATTRACT/downloads/spydersilk-0.03.tar.gz)
- Install pip (under Ubuntu: sudo apt-get install python-pip)
- Install Spydersilk: sudo pip install spydersilk-0.03.tar.gz
- To test that it works, type "cd $ATTRACTGUI; python -c 'import spyder, attractmodel'"

11. ATTRACT-EM GPU routines
These require the PyCUDA library. Install pip (see above), then PyCUDA (under Ubuntu: sudo pip install pycuda)
You will also need to install jinja2 and cffi (both with pip)
Finally, you need to go to the attract-em-gpu directory, and run axsym_build.py