bio-pyvol 1.1.1

a PyMOL plugin for visualization, comparison, and volume calculation for protein drug-binding sites

PyVOL

PyVOL is a python library for identifying protein binding pockets, partitioning them into sub-pockets, and calculating their volumes. While PyVOL can be used to identify new binding pockets, there are many algorithms that have been optimized for that task. PyVOL is intended to be used to describe features of known binding pockets. PyVOL can be run as a PyMOL plugin, as an imported python library, or as a commandline program. Visualization of results is exclusively supported through PyMOL.

While the API is not guaranteed to be stable at this point, it is unlikely to change.

Basic PyMOL Installation

PyVOL can be installed into PyMOL by using the plugin manager to install the zip file downloaded from:

https://github.com/rhs2132/pyvol/blob/master/pyvol_plugin.zip

This adds a menu option under plugins "Install PyVOL." Clicking this and selecting install will download PyVOL and all its dependencies. On MacOS and Linux, this should be a complete installation. Windows currently requires independent installation of MSMS. PyVOL will be available to run once PyMOL is restarted.

Basic Manual Installation

PyVOL minimally requires biopython, msms, numpy, pandas, scipy, scikit-learn, and trimesh in order to run. PyVOL is available for manual installation from github or from PyPI.

pip install bio-pyvol

MSMS must be installed separately. The bioconda channel provides a version for Linux and OSX:

conda install -c bioconda msms

Otherwise MSMS must be installed manually by downloading it from MGLTools and adding it to the path. PyMOL distributions from Schrodinger have MSMS included; however, it must still be added to the path manually. The executable is located at:

<pymol_root_dir>/pkgs/msms-2.6.1-2/bin/msms

As mentioned before, visualization relies on PyMOL 2.0+. Once PyVOL is installed in the python environment used by PyMOL, the script can be installed by using the plugin manager to install the file "pyvol_plugin.zip".

Detailed PyMOL Installation

Installing into PyMOL can be unexpectedly difficult due to some quirks in PyMOL's dependency management. Specifically, some versions of PyMOL 2.0+ use both pip and conda to manage dependencies which can lead to odd conflicts. The libraries managed by pip seem to take precedence over those by conda, so preferentially use the pip installation path. The pip and conda executables packaged with PyMOL are located at:

<pymol_root_dir>/bin/conda
<pymol_root_dir>/bin/pip

Quick Start

From within PyMOL, the simplest binding pocket calculation is simply run with:

pocket protein_selection

The two parameters that most dramatically affect calculations are the maximum and minimum radii used to respectively define the exterior surface of the protein and the boundary of the binding pocket itself. In practice, the minimum radius does not need to be changed as its default (1.4) is broadly useful. The maximum radius does often need to be adjust to find a suitable value using the max_rad paramter:

pocket protein_selection, min_rad=1.4, max_rad=3.4

Basic Usage

Pocket Specification

PyVOL by default recognizes the largest pocket and returns the volume and geometry for it. However, manual identification of the pocket of interest is generally preferable. This can be done through specification of a ligand, a residue, or a coordinate. If a specification is given, the mode is changed to specific by default.

Default behavior:

pocket protein_selection, mode=largest

Ligand specification:

pocket protein_selection, mode=specific, ligand=ligand_selection
pocket protein_selection, ligand=ligand_selection

Residue specification:

pocket protein_selection, mode=specific, resid=A15
pocket protein_selection, resid=A15
pocket protein_selection, mode=specific, residue=residue_selection
pocket protein_selection, residue=residue_selection

where the resid is written as . If there is only one chain in the selection, the chain ID can be excluded.

Coordinate specification:

pocket protein_selection, mode=specific, pocket_coordinate="5.0 10.0 15.0"
pocket protein_selection, pocket_coordinate="5.0 10.0 15.0"

where the coordinate is provided as three floats separated by spaces and bounded by quotation marks.

Alternatively, PyVOL can return the surfaces and volumes for all pockets above a minimum volume that are identified. By default, this volume cutoff is set at 200 A^3.

pocket protein_selection, mode=all, minimum_volume=200

Extra Ligand Options

When a ligand is provided, the atoms of the ligand can be used to identify both minimum and maximum extents of the calculated binding pocket. To include the volume of the ligand in the pocket volume (useful for when the ligand extends into bulk solvent), use the lig_incl_rad parameter:

pocket protein_selection, ligand=ligand_selection, lig_incl_rad=0.0

where the value of lig_incl_rad is added to the Van der Waals radii of each atom in the ligand selection when calculating the exterior surface of the protein.

The atoms of the ligand can also be used to define a maximum boundary to the calculated pocket by specifying the lig_excl_rad parameter:

pocket protein_selection, ligand=ligand_selection, lig_excl_rad=2.0

where the value of lig_excl_rad is added to the Van der Waals radii of each atom in the ligand selection when calculating the exterior surface of the protein.

Sub-pocket Partitioning

Sub-partitioning is enabled by setting the subdivide paramter to True:

pocket protein_selection, subdivide=True

Parameters controlling the number of sub-pockets identified generally perform well using defaults; however, they can be easily adjusted as needed. The two most important parameters are the minimum radius of the largest sphere in each sub-pocket (this excludes small sub-pockets) and the maximum number of clusters:

pocket protein_selection, subdivide=True, min_subpocket_rad=1.7, max_clusters=10

If the number of clusters must be reduced, sub-pockets are merged on the basis of connectivity between the defining sets of tangent spheres. Practically, sub-pockets with a greater surface area boundary are merged first.

Display and Output Options

By default, PyVOL simply writes volumes to STDOUT and, when invoked through PyMOL, displays pocket boundaries as semi-translucent surfaces. This behavior can be extensively customized.

The output name for all computed PyMOL objects and the base filename for any output files can be specified using the prefix option:

pocket protein_selection, prefix=favprot

PyVOL can also write the input and output files to a directory if given an output directory. In this case it writes out the input protein and ligand structures, a csv report of all calcuated volumes, and paired csv/obj files containing tangent sphere collections and 3D triangulated mesh files respectively.

pocket protein_selection, output_dir=best_out_dir

Calculated surfaces can be visualized in three different ways by setting the display_mode parameter. The following three commands set the output as a solid surface with transparency, a wireframe mesh, and a collection of spheres. Color is set with the color parameter and transparency (when applicable) with the alpha parameter:

pocket protein_selection, display_mode=solid, alpha=0.85, color=skyblue
pocket protein_selection, display_mode=mesh, color=red
pocket protein_selection, display_mode=spheres, color=firebrick

where alpha is [0, 1.0] and the color is any color defined within pymol. The presets should generally be sufficient, but custom colors can be chosen using the commands given on the PyMOL wiki.

Command-line Interface

PyVOL can also be run from the command-line. If installed using pip, a "pyvol" entry point should be automatically installed and made available on the path. Otherwise, manual invocation of pyvol/main.py should work. From the command-line, PyVOL is run with a standard configuration file.

python -m pyvol <input_parameters.cfg>

A template configuration file with default values supplied can be generated using:

python -m pyvol -t <output_template.cfg>

Currently, PyVOL does not output any information to stdout when run this way. So if an output directory is not provided, there is no easy way to retrieve the results.