Python-based Computation of Protein Electric Field Topology, built for high-throughput accelerated calculations of several electrostatic properties of enzyme active sites from simulations.
To cite your use of pycpet, please use the following: Ajmera, P., Vargas, S., Chaturvedi, S., Hennefarth, M. & Alexandrova, A. PyCPET - Computing Heterogeneous 3D Protein Electric Fields and Their Dynamics. ChemRxiv (2024). doi:10.26434/chemrxiv-2024-r3thp
System requirements:
- gcc (to compile C-shared libraries)
- anaconda (preferred, not required)
Follow these steps to install PyCPET
- Clone the repo:
git clone https://github.com/pujanajmera/CPET-python.git
- (Highly recommended but not required) Install the pre-made conda environment CPET_ENV in a path directory:
conda env create -f CPET_ENV.yml -p PATH_TO_ENVIRONMENT_LOCATION
If you do not create the environment/this environment file doesn't work for you, you will have to install the required python packages yourself (see CPET_ENV.yml)
- In the current directory, do the following (if you are using CPET_ENV, make sure to be in the environment when running this step):
pip install -e .
From here, the executable cpet.py will be available in your path (or in your environment if using conda)
- To compile the C math modules, go to CPET/utils and run:
cd CPET/utils gcc -fopenmp math_module.c -o math_module.so -shared -fPIC -O3 -march=native -funroll-loops -ffast-math
- When running, we advise you set the following:
CPU multithreading: export OMP_NUM_THREADS=1 GPU-accelerated code: export CUDA_VISIBILE_DEVICES=N (where N is the GPU number, only needed for non-HPC setups)
These are the following available features, and their corresponding options file 'method' keywords:
- Computing point electric fields: 'point_field'
- Computing point electric field magnitudes: 'point_mag'
- Computing 3-D electric fields: 'volume'
- Computing 3-D electrostatic potentials: 'volume_ESP'
- Computing 3-D distribution of streamlines: 'topo' (CPU, default) and 'topo_GPU' (GPU)
- Clustering by distribution of streamlines: 'cluster'
- Clustering by 3-D electric field (tensor decomp): 'cluster_volume_tensor' IN BETA
- Clustering by 3-D electrostatic potential: 'cluster_volume_ESP_tensor' IN BETA
- Visualizing 3-D fields: 'visualize_fields'
- PCA on 3-D fields, for a single set of data: 'pca' IN BETA
- PCA on 3-D fields, for a full comparative analysis between multiple sets of field data (e.g. variants): 'pca_compare' IN BETA
- Finding if any atoms are intruding the field volume: 'box_check'
Several examples are in the examples directory. Most of these are designed for single calculations, but can be extended to high-throughput with almost no changes.
For features unavailable from the cpet.py script mentioned above, we offer scripts in source/scripts for the following. Please note that these are not rigorously tested for all cases, but showcase the scripting ability of the pycpet library:
- residue_breakdown_analysis.py: Residue contribution to topology over dynamics, ranked. Requires completed topology calculations for an MD
- tensor_based_cluster_double.py: Electrostatic potential/electric field-based clustering for two sets of electric fields/electrostatic potentials. Assumes that all field calculations have been completed for both sets of directories.
Most of the code here requires well-formated PDB or PQR files. The formatting is as follows (see io.py for more details):
PDB:
- Up to and including charge information, follow standard formatting: https://www.cgl.ucsf.edu/chimerax/docs/user/formats/pdbintro.html
- For charge information, charges must be in columns 55-64 of each line, inclusive
