The Minimum Distance Matrix Representation (MDMR) platform for use in within Jupyter notebooks. This method calculates distance matrices for a set of structures and run analyses that classify the protein conformational and ligand binding mode ensembles. Further analyses identifies important intra-protein interactions. This is the accompanying GitHub for the paper Can Deep Learning Blind Docking Methods be used to Predict Allosteric Compounds?
- Free software: MIT license
- Calculates minimum distance matrices on receptor conformations or ligand binding modes
- Runs PCA and HDBSCAN on distance matrices
- Calculates nSMD and accompanying histograms
- Uses altair for interactive plots and py3Dmol for visualizing structures within the notebooks
- Code to re-run docking benchmarks as described in the paper
Create conda environment for this package
conda create --name MDMR python=3.9 conda activate MDMR conda install mdtraj -c conda-forge git clone https://github.com/echen1214/Min_Dist_Matrix_Rep cd Min_Dist_Matrix_Rep pip install -e .
This package is intended to be used with Jupyter notebooks. You can use the conda environment you just created within the Jupyter notebook
pip install jupyterlab pip install notebook conda install -c conda-forge nb_conda_kernels
Download and install a ClustalOmega binary into a local directory and symlink the binary to the command line. For example Mac users can do
wget http://www.clustal.org/omega/clustal-omega-1.2.3-macosx
mv clustal-omega-1.2.3-macosx clustalo
sudo chmod u+x clustalo
sudo ln -s /{path}/{to}/clustalo /usr/local/bin/
## this will be different with zsh shell
Similarly, a Windows user can do
wget http://www.clustal.org/omega/clustal-omega-1.2.2-win64.zip -outfile clustal.zip tar -xf clustal.zip cd clustal-omega-1.2.2-win64 mklink clustalo .\clustalo.exe set PATH=%PATH%;%cd%
You can run tutorial/Tutorial1_CDK2.ipynb with the data provided straight from the GitHub repository. This tutorial walks you through the downloading, processing, and calculating the distance matrices of CDK2 structures. Then you perform downstream analyses with PCA and HDBSCAN of the receptor-only and ligand-only plots, and nSMD calculations and histograms.
The self- and cross-docking benchmarking analyses can be found tutorial/Tutorial2_self_cross_docking.ipynb. To run the analyses first download and untar the poses zenodo_tar.tar.gz from the Zenodo database. The tutorial then performs the calculations of the main self- and cross-docking results of the paper.
python mdtraj numpy scikit-learn matplotlib biopython prody py3Dmol ipykernel altair rcsbsearchapi anytree pypdb spyrmsd meeko AlphaSpace2
This package was created with Cookiecutter and the audreyr/cookiecutter-pypackage project template.