3DFSC Program Suite Version 3.0 (16 March 2018)
This is the most recent version of 3DFSC, by Philip Baldwin, Yong Zi Tan, and Dmitry Lyumkis.
GPU code and Conda environment by Carl Negro.
3DFSC Program Suite requires Anaconda 3 to run, and UCSF Chimera to visualize the outputs.
- Navigate to the directory where you want to install the 3DFSC Program Suite.
git clone https://github.com/nysbc/Anisotropy.gitto clone a copy of the program from the Github repository. Make sure you have git installed properly or else this will fail.
cd Anisotropyto enter the directory.
- Run the init script by executing the command
./init.sh. This will create the necessary Anaconda3 environment, auto-detect your repository path, and generate run3DFSC executable scripts for bash, sh, and csh terminal shells. This will only work if Anaconda 3 is installed and Anaconda's Python interpreter is on your PATH variable.
- Navigate to the directory containing your maps.
- Copy the run3DFSC.bash, .csh or .sh script to this directory.
- To view the 3DFSC parameters, access the help info like
- Execute the run3DFSC script with the appropriate parameters. It can take from minutes for small maps up to hours for extremely large box sizes (we have tested 600^3). Progress bars will help indicate the state of processing.
3DFSC now has GPU support through Numba for faster execution (typically ~10x faster than the CPU implementation). This requires that CUDA is installed correctly. See http://www.nvidia.com/Download/index.aspx.
To make use of a GPU, simply append the
--gpu flag as a parameter to your ./run3DFSC.sh script.
You can select which GPU to use for processing with the
--gpu_id flag. E.g.
--gpu_id=2 to run 3DFSC on the GPU with index number 2. The program currently only allows a single GPU to be used at once; this may change in future versions.
To see a list of available GPU's and corresponding indices, run
nvidia-smi. If you are unable to run
nvidia-smi, check to make sure you have CUDA installed correctly. See http://www.nvidia.com/Download/index.aspx.
Note that GPU memory is limited, so that processing jobs with large box sizes will fail.
Example: Haemagglutinin Trimer with Preferred Orientation Collected at Tilts
- Go the Example directory
../run3DFSC.csh --halfmap1=T40_map1_Masked_144.mrc --halfmap2=T40_map2_Masked_144.mrc --fullmap=130K-T40.mrc --apix=1.31 --ThreeDFSC=T40-3DFSCto run the Example. It should take no longer than 2 minutes (~15-20 seconds with a Pascal-generation GPU).
- The folder already contains pre-calculated results as well as a log of a successful run (
Questions and feedback welcomed, and should be sent to email@example.com, firstname.lastname@example.org and email@example.com.