Feshdock is focused on protein-protein docking method based on the Fast Fourier Transform (FFT) and evolutionary computation algorithms. Feshdock leverages the FFT to globally scan potential binding positions, which are then partitioned into multiple clusters using the K-means method. These clusters are locally optimized by a swarm intelligence algorithm in a divide-and-conquer manner, accompanied by an anisotropic network model considering protein flexibility. Finally, hierarchical clustering is employed to determine the final candidate complex conformations.
Feshdock is recommended to be compiled and executed in the Linux OS, it has not been tested on Microsoft Windows and may bave some issues.
Create a virtual environment and install some necessary dependencies first:
-
conda create -n feshdock python=3.9 -
conda activate feshdock
Feshdock has following important dependencies:
- biopython==1.79
- Cython==3.0.10
- freesasa==2.2.0
- nvidia-nccl-cu12==2.22.3
- pyFFTW==0.13.1
- scikit-learn==0.24.1
- transforms3d==0.4.1
all dependencies are listed in the requirements.txt (pip install -r requirements.txt).
Then compile the Scoring module. Enter the directory feshdock/scoring/dfire/cython and execute:
-
cd feshdock/scoring/dfire/cython -
python setup.py build_ext --inplace
Remember that all executables need to be authorized by chmod +x [executable file].
Place the receptor protein, ligand protein, and native comple structure files to be docked in the data/ directory, then execute the following command for docking:
python run_feshdock.py
After docking, result will be saved in data/fina_models/.