A PyMOL plugin with accompanying Docker image for kinase inhibitor binding and affinity prediction
KinaseDocker² is a computational tool that implements fully automated docking and scoring. The tool allows for docking in either AutoDock VinaGPU or DiffDock and subsequent scoring by a Deep Neural Network that has been trained on kinase-inhibitor docking poses. This tool can both be installed as a PyMOL plugin and used through the CLI.
In the backend, it uses a Docker image to run the GPU-accelerated VinaGPU, DiffDock and PyTorch DNN implementation. The instructions below assume you have a working GPU-enable Docker installation on your system. Refer to guides such as https://docs.nvidia.com/datacenter/cloud-native/container-toolkit/latest/install-guide.html for detailed Docker installation instructions.
- Download the KinaseDocker.yml file
- Create environment
conda env create -f KinaseDocker.yml- Activate environment
conda activate KinaseDockerOr, if you really want to install it in an existing environment:
- Get PyMOL
conda install -c conda-forge PyMOL-open-source- Install dependencies
conda install anaconda::h5py
pip install meeko==0.3.3 scipy docker pandas rdkit- Download and load the docker image
docker pull apajanssen/kinasedocker2- Download kinasedocker_plugin.zip
- Run PyMOL
pymol- Go to Plugin > Plugin manager > Install New Plugin
- Click Choose file... and select the kinasedocker_plugin.zip
- Press Ok a bunch of times
- Create environment
conda create -n MY_ENV python=3.9 - Activate environment
conda activate MY_ENV- Install dependecies
pip install meeko==0.3.3 scipy docker pandas rdkit- Download the docker image: vina_diffdock_dnn.tar
- Load the image
docker load -i vina_diffdock_dnn.tar- Download the files pipeline.py and kinase_data.csv
- Run the pipeline
python pipeline.py --help