PowerFit is a Python package and simple command-line program to automatically fit high-resolution atomic structures in cryo-EM densities. To this end it performs a full-exhaustive 6-dimensional cross-correlation search between the atomic structure and the density. It takes as input an atomic structure in PDB-format and a cryo-EM density with its resolution; and outputs positions and rotations of the atomic structure corresponding to high correlation values. PowerFit uses the local cross-correlation function as its base score. The score can optionally be enhanced by a Laplace pre-filter and/or a core-weighted version to minimize overlapping densities from neighboring subunits. It can further be hardware-accelerated by leveraging multi-core CPU machines out of the box or by GPU via the OpenCL framework. PowerFit is Free Software and has been succesfully installed and used on Linux and MacOSX machines.
Minimal requirements for the CPU version:
- Python3.10 or greater
- NumPy 1.8+
- SciPy
- GCC (or another C-compiler)
- FFTW3
- pyFFTW 0.10+
To offload computations to a discrete or integrated* GPU the following is also required
- OpenCL1.1+
- pyopencl
- pyvkfft
Recommended for installation
- git
- pip
* Integrated graphics on CPUs are able to signficantly outperform the native CPU implementation in some cases. This is mostly applicable to Intel devices, see the section tested platfoms.
If you already have fulfilled the requirements, the installation should be as easy as opening up a shell and typing
# To run on CPU
pip install powerfit-em
# To run on GPU
pip install powerfit-em[opencl]If you are starting from a clean system, follow the instructions for your particular operating system as described below, they should get you up and running in no time.
If you do not have system admin rights, you likely cannot compile pyvkfft locally.
However, by installing powerfit in a conda environment, you can still do computations
on GPU. If you are on a Linux system and have Conda or Mamba available, follow
these instructions;
Steps for running on GPU with Conda
For AMD or NVIDIA GPUs you can run the following command. Note that this relies
on OpenCL drivers being available system wide (under /etc/OpenCL/vendors/).
conda create -n powerfit -c conda-forge python=3.12 ocl-icd ocl-icd-system pyopencl pyvkfft
conda activate powerfit
pip install powerfit-em[opencl]On Intel integrated graphics you can use the following command. This includes the OpenCL runtime and does not rely on your system setup:
conda create -n powerfit -c conda-forge python=3.12 ocl-icd intel-compute-runtime pyopencl pyvkfft
conda activate powerfit
pip install powerfit-em[opencl]Some older Intel processors might need to use intel-opencl-rt instead of intel-compute-runtime.
After installation, you can check that the OpenCL installation is working by running
python -c 'import pyopencl as cl;from pyvkfft.fft import rfftn; ps=cl.get_platforms();print(ps);print(ps[0].get_devices())'Powerfit can be run in a Docker container.
Install docker by following the instructions.
Linux systems usually already include a Python3.10 or greater distribution. First make sure the Python header files, pip and git are available by opening up a terminal and typing for Debian and Ubuntu systems
sudo apt update
sudo apt install python3-dev python3-pip git build-essentialIf you are working on Fedora, this should be replaced by
sudo yum install python3-devel python3-pip git development-c development-toolsSteps for running on GPU
If you want to use the GPU version of PowerFit, you need to install the drivers for your GPU.
After installing the drivers, you need to install the OpenCL development libraries. For Debian/Ubuntu, this can be done by running
sudo apt install ocl-icd-opencl-dev ocl-icd-libopencl1For Fedora, this can be done by running
sudo dnf install opencl-headers ocl-icd-develInstall pyvkfft, a Python wrapper for the VkFFT library, using
pip install pyvkfftCheck that the OpenCL installation is working by running
python -c 'import pyopencl as cl;from pyvkfft.fft import rfftn; ps=cl.get_platforms();print(ps);print(ps[0].get_devices())'
# Should print the name of your GPUYour system is now prepared, follow the general instructions above to install PowerFit.
First install git by following the instructions on their website, or using a package manager such as brew
brew install gitNext install pip, the Python package manager, by following the installation instructions on the website or open a terminal and type
python -m ensurepip --upgradeTo get faster score calculation, install the pyFTTW Python package in your conda environment
with conda install -c conda-forge pyfftw.
Follow the general instructions above to install PowerFit.
First install git for Windows, as it comes with a handy bash shell. Go to git-scm, download git and install it. Next, install a Python distribution such as Anaconda. After installation, open up the bash shell shipped with git and follow the general instructions written above.
After installing PowerFit the command line tool powerfit should be at your disposal. The general pattern to invoke powerfit is
powerfit <map> <resolution> <pdb>where <map> is a density map in CCP4 or MRC-format, <resolution> is the
resolution of the map in ångstrom, and <pdb> is an atomic model in the
PDB-format. This performs a 10° rotational search using the local
cross-correlation score on a single CPU-core. During the search, powerfit
will update you about the progress of the search if you are using it
interactively in the shell.
Usage in Docker
The Docker images of powerfit are available in the GitHub Container Registry.
Running PowerFit in a Docker container with data located at
a hypothetical /path/to/data on your machine can be done as follows
docker run --rm -ti --user $(id -u):$(id -g) \
-v /path/to/data:/data ghcr.io/haddocking/powerfit:v3.1.0 \
/data/<map> <resolution> /data/<pdb> \
-d /data/<results-dir>For <map>, <pdb>, <results-dir> use paths relative to /path/to/data.
To run tutorial example use
# cd into powerfit-tutorial repo
docker run --rm -ti --user $(id -u):$(id -g) \
-v $PWD:/data ghcr.io/haddocking/powerfit:v3.1.0 \
/data/ribosome-KsgA.map 13 /data/KsgA.pdb \
-a 20 -p 2 -l -d /data/run-KsgA-dockerTo run on NVIDIA GPU using NVIDIA container toolkit use
docker run --rm -ti \
--runtime=nvidia --gpus all -v /etc/OpenCL:/etc/OpenCL \
-v $PWD:/data ghcr.io/haddocking/powerfit:v3.1.0 \
/data/ribosome-KsgA.map 13 /data/KsgA.pdb \
-a 20 -l -d /data/run-KsgA-docker-nv --gpuTo run on Intel integrated graphics use
docker run --rm -ti \
--device=/dev/dri \
-v $PWD:/data ghcr.io/haddocking/powerfit:v3.1.0 \
/data/ribosome-KsgA.map 13 /data/KsgA.pdb \
-a 20 -l -d /data/run-KsgA-docker-nv --gpuTo run on AMD GPU use
sudo docker run --rm -ti \
--device=/dev/kfd --device=/dev/dri \
--security-opt seccomp=unconfined \
--group-add video --ipc=host \
-v $PWD:/data ghcr.io/haddocking/powerfit-rocm:v3.1.0 \
/data/ribosome-KsgA.map 13 /data/KsgA.pdb \
-a 20 -l -d /data/run-KsgA-docker-amd --gpuFirst, to see all options and their descriptions type
powerfit --helpThe information should explain all options decently. In addtion, here are some examples for common operations.
To perform a search with an approximate 24° rotational sampling interval with laplace pre-filtering and core-weighted scoring function using 1 CPU
powerfit <map> <resolution> <pdb> -a 24To use multiple CPU cores without laplace pre-filter and 5° rotational interval
powerfit <map> <resolution> <pdb> -p 4 --no-laplace -a 5To off-load computations to the GPU and do not use the core-weighted scoring function and write out the top 15 solutions
powerfit <map> <resolution> <pdb> -g --no-core-weighted -n 15Note that all options can be combined except for the -g and -p flag:
calculations are either performed on the CPU or GPU.
To run on GPU
powerfit <map> <resolution> <pdb> --gpu
...
Using GPU-accelerated search.
...When the search is finished, several output files are created
- fit_N.pdb: the top N best fits.
- solutions.out: all the non-redundant solutions found, ordered by their correlation score. The first column shows the rank, column 2 the correlation score, column 3 and 4 the Fisher z-score and the number of standard deviations (see N. Volkmann 2009, and Van Zundert and Bonvin 2016); column 5 to 7 are the x, y and z coordinate of the center of the chain; column 8 to 17 are the rotation matrix values.
- lcc.mrc: a cross-correlation map, showing at each grid position the highest correlation score found during the rotational search.
- powerfit.log: a log file, including the input parameters with date and timing information.
- report.html and state.mvsj: an HTML report and its MolViewSpec with interactive 3D visualization of the best fits.
Only written if the
--report --delimiter ,arguments are passed.
If this software was useful to your research, please cite us
G.C.P. van Zundert and A.M.J.J. Bonvin. Fast and sensitive rigid-body fitting into cryo-EM density maps with PowerFit. AIMS Biophysics 2, 73-87 (2015) https://doi.org/10.3934/biophy.2015.2.73.
For the use of image-pyramids and reliability measures for fitting, please cite
G.C.P van Zundert and A.M.J.J. Bonvin. Defining the limits and reliability of rigid-body fitting in cryo-EM maps using multi-scale image pyramids. J. Struct. Biol. 195, 252-258 (2016) https://doi.org/10.1016/j.jsb.2016.06.011.
If you used PowerFit v1, please cite software with https://doi.org/10.5281/zenodo.1037227. For version 2 or higher, please cite software with https://doi.org/10.5281/zenodo.14185749.
Apache License Version 2.0
The elements.py module is licensed under MIT License (see header). Copyright (c) 2005-2015, Christoph Gohlke
| Operating System | CPU single | CPU multi | GPU |
|---|---|---|---|
| Linux | Yes | Yes | Yes |
| MacOSX | Yes | Yes | No |
| Windows | Yes | Fail | No |
The GPU version has been successfully tested on Linux and with a Docker container for the following devices;
- NVIDIA GeForce GTX 1050 Ti
- NVIDIA GeForce RTX 4070
- AMD Radeon RX 7800 XT
- AMD Radeon RX 7900 XTX
- Intel Iris Xe Graphics (on a Core i7-1185G7)
The integrated graphics of AMD Ryzen CPUs do not officially support OpenCL. If they do seem available in PyOpenCL be aware that this may lead to incorrect results.
To contribute to PowerFit, see CONTRIBUTING.md.