Skip to content
/ DockBox Public

Python package to facilitate the use of popular docking software

License

Notifications You must be signed in to change notification settings

jp43/DockBox

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

https://github.com/jp43/DockBox/raw/master/logo.png

DockBox is a python wrapper library designed to facilitate the use of standard docking programs either alone or in combination. In addition, DockBox enables to rescore the generated docking poses with different popular scoring algorithms as well as to analyze the results using different consensus docking/scoring strategies.

List of programs supported by DockBox

None of the above docking or scoring programs are included in the current repository. Hence the user eager to test a docking/scoring program with DockBox should first install the program separately on the same system DockBox is installed.

To make a docking or scoring program usable by DockBox, make sure all the executable files required for each program can be directly run on the command line from any location, i.e., are in your PATH (see section executable files required for each program to find out the executables needed to run each docking/scoring programs). For MOE, additional environment variable, i.e, MOE, must be set up.

The following are the minimal requirements to install the DockBox module

  • python version 2.6 or 2.7
  • virtualenv version 1.11 or later

The easiest way to install DockBox is to create a virtual environment. In this way, DockBox and its dependencies can easily be installed in user-space without clashing with potentially incompatible system-wide packages.

Once virtualenv has been properly installed, simply type (and press the return key)

virtualenv env

on the command line followed by

source env/bin/activate

to activate the virtual environment (do not forget to activate your environment every time you log into a new shell environment).

Finally, the DockBox package can be set up by going in DockBox installation directory and typing:

python setup.py install

Installation is complete!

Any docking/scoring software intended to be used by DockBox should be installed separetely and should work as a standalone program. In addition, make sure the following exectuables are in your PATH, depending on which docking/scoring software should be used:

  • AutoDock 4

    • autodock4
    • autogrid4
    • obabel
    • prepare_dpf4.py
    • prepare_gpf4.py
    • prepare_ligand4.py
    • prepare_receptor4.py

    The autodock4 and autogrid4 executables can be installed by installing AutoDock4. All the .py executable files are included in the MGLTools package. It is important to not move those executable to different locations outside of the MGLTools package as DockBox uses those locations to deduce additional MGLTools python modules (PyBabel, MolKit). The babel routine can be set by installing OpenBabel from it official website.

  • AutoDock Vina

    • obabel
    • prepare_ligand4.py
    • prepare_receptor4.py
    • vina

    The vina executable is available within the AutoDock Vina package. See AutoDock 4 section for other executables.

  • DOCK 6

    • chimera
    • dms
    • dock6
    • grid
    • showbox
    • sphere_selector
    • sphgen_cpp

    The chimera command is accessible when installing CHIMERA. The dock6, grid, showbox and sphere_selector commands are all part of the DOCK 6 package. The dms program can be installed seperately from here while sphgen_cpp can be downloaded from the following url.

  • DSX

    • dsx

    The dsx-like routines can be downloaded from the official link. Note that DockBox assumes DSX is run via an executable file called dsx. Therefore, once you selected the appropriate routine to be run (e.g., dsx_linux_64.lnx for linux systems), create a symbolic link via ln. For example, assuming you are located in the same directory as dsx_linux_64.lnx, you can use the following command:

ln -s dsx_linux_64.lnx dsx
  • MOE

    • moebatch

    Can be downloaded over there. Make sure the MOE environment variable (pointing towards MOE's installation directory) has been correctly exported. Note that a valid license of MOE is required.

Note: In case rescoring is enabled (rescoring=yes in the configuration file, see below), it is recommended to also enable minimization of the poses (minimize=yes in the configuration file). In this case, AmberTools (serial version) 14, 15, 16 or 17 is required for minimization. Make sure main AMBER executables are accessible from the PATH variable including sander, tleap, antechamber and parmchk which are used by DockBox.

The DockBox package contains two main routines: rundbx and extract_dbx_best_poses. The former is intended to be used solely for docking and rescoring while the latter enables to analyze the results and to select the best pose(s) from a combination of scores or among different consensus docking schemes.

rundbx is used to dock a ligand to a protein structure and possibly minimize and rescore the output poses. When typing "rundbx -h" on the command line, the following help message will pop up:

usage: rundbx [-h] -l INPUT_FILE_L -r INPUT_FILE_R -f CONFIG_FILE
              [-prepare_only] [-rescore_only]

rundbx : dock and rescore with multiple programs -------- Requires one file
for the ligand (1 struct.) and one file for the receptor (1 struct.)

optional arguments:
  -h, --help       show this help message and exit
  -l INPUT_FILE_L  Ligand coordinate file(s): .mol2
  -r INPUT_FILE_R  Receptor coordinate file(s): .pdb
  -f CONFIG_FILE   config file containing docking parameters
  -prepare_only    Only prepare scripts for docking (does not run docking)
  -rescore_only    Run rescoring only
  • Inputs
    • -l INPUT_FILE_L: .mol2 file containing the coordinates of the ligand (only one structure allowed)
    • -r INPUT_FILE_R: .pdb file containing the receptor coordinates (only one structure allowed)
    • -f CONFIG_FILE: .ini configuration file containing the docking parameters (see the section preparing the rundbx configuration file)
  • Options
    • -prepare_only: generate all docking folders and scripts needed to run each docking program separately. Does actually not run docking.
    • -rescore_only: option used to perform the rescoring step only. Using this option implies that you have already run rundbx and generated a poses folder in the current directory. If a rescoring folder already exists as an output of a previous rundbx run, every data generated previously by rescoring with the same scoring functions as the current ones will be overwritten while data generated with scoring functions different from the current ones will be kept.

extract_dbx_best_poses is the routine used to analyze the results and select the best docking poses from the outputs of the rundbx command. When typing "extract_dbx_best_poses -h" on the command line, the following help message will pop up:

usage: extract_dbx_best_poses [-h] [-all-targets] [-all-isomers] [-csv FILE]
                            [-cutoff RMSD_VALUE] [-d PRGM1 [PRGM2 ...]]
                            [-dirs DIR1 [DIR2 ...]] [-r DIRECTORY NAME]
                            [-s FUNC [FUNC ...] | -cd PRGM [PRGM ...] |
                            -sbcd FUNC [FUNC ...]]

Extract best docking poses after rundbx finished.

optional arguments:
-h, --help            show this help message and exit
-all-targets          Select best poses over all the targets. If not
                      specified, extract best pose separately for each
                      target. A "lig/target/isomer" architecture of the
                      folders is assumed
-all-isomers          Select best poses over all the isomers. If not
                      specified, extract best pose separately for every
                      isomer. A "lig/target/isomer" architecture of the
                      folders is assumed
-csv FILE             .csv filename with compounds. Used to add names of
                      compounds. Default: none
-cutoff RMSD_VALUE    RMSD cutoff used for consensus docking or score-based
                      consensus docking. Default: 2.0 A
-d PRGM1 [PRGM2 ...]  Docking programs (instances) to be considered when
                      extracting best poses
-dirs DIR1 [DIR2 ...]
                      Directories considered for analysis. Should contain a
                      folder called "poses". Default: curr. dir
-r DIRECTORY NAME     Name of results directory. Default: results
-s FUNC [FUNC ...]    Scoring functions used to extract the best pose
                      (combination of scores)
-cd PRGM [PRGM ...]   Docking programs used for standard consensus docking
-sbcd FUNC [FUNC ...]
                      Scoring functions used for score-based consensus
                      docking

The rundbx routine allows the user to dock and rescore a ligand to a protein target using multiple docking and scoring functions. Running rundbx is fairly simple as it requires only three input files, namely a PDB file including the protein structure to dock on (-r flag), a file with Tripos Mol2 format containing a 3D structure of the ligand (1 structure per file, -l flag) and an INI configuration file (-f flag) which contains all the options related to docking and/or rescoring (see section Preparing the INI configuration file).

When finished correcly, a rundbx job should have created a folder called poses containing all the poses generated by the different docking programs as specified in the INI configuration file. Each pose is provided in a file with .mol2 format named lig-<index>.mol2, where <index> is the index of the pose. Within the poses folder, a file called info.dat can also be found. The file contains information relative to each docking program/site combination specified in the INI file, including the number of poses generated and the index of the first pose generated for that combination.

Below is an example of an info.dat file obtained when docking was performed with Autodock, Autodock Vina and DOCK 6 on the same binding site:

#1,28
program,nposes,firstidx,site
autodock,10,1,
vina,10,11,
dock,7,21,

showing that a total of 27 poses were generated (28 minus 1). 10 poses were generated with Autodock, namely, poses from index 1 to 10, 10 were generated with Autodock Vina, i.e., poses from index 11 to 20, and 7 were generated with DOCK 6, poses from index 21 to 27. No label for the binding site was specified as docking was performed on the same site.

Other outputs of the rundbx command are folders created for every docking program/site combination specified in the INI file, which contain the docking poses, the docking scores (obtained with docking) and intermediate files generated by the docking software. For example, if Autodock and Autodock Vina were used to dock on three different binding sites called site1, site2 and site3 (see section Preparing the INI configuration file), then a total of six folders named autodock.site1, autodock.site2, autodock.site3, vina.site1, vina.site2 and vina.site3, should have been created.

Finally, if the rescoring option was enabled in the INI file, a folder called rescoring should have been created as well, containing file(s) named <program>.score, where <program> is the name of each program used for rescoring.

Besides one MOL2 file containing the ligand structure (-l flag) and one PDB file containing the receptor structure (-r flag), running rundbx requires a configuration file (-f flag) that specifies all the parameters needed for the docking procedure.

The rundbx configuration file should be a INI file (https://en.wikipedia.org/wiki/INI_file), i.e., the file should be split in sections, each section name appearing on a line by itself, in square brackets ("[" and "]"). Each section contains a certain number of keys which refer to specific options used; all keys after the section declaration are associated with that section. Finally, every key should have a name (option name) and a value (option value), delimited by an equals sign (=).

Below is an example of configuration file used to dock on two binding sites and rescore with DrugScoreX (dsx), Autodock and Autodock Vina.

[DOCKING]
site = site1, site2
program = autodock, vina, dock
rescoring = yes
minimize = yes
cleanup = yes

[RESCORING]
program = dsx, autodock, vina

[DSX]
pot_dir = /pmshare/jordane/CSD_potentials/DSX_CSD_Potentials_v0511/csd_pot_0511/
other_flags = -T0 1.0 -T1 1.0 -T3 1.0 -j

[AUTODOCK]
ga_run = 20
spacing = 0.4

[VINA]
num_modes = 20

[DOCK]
nposes = 20

[SITE1]
center = 75.5, 80.0, 31.0
boxsize = 40.0, 40.0, 40.0

[SITE2]
center = 75.5, 40.0, 50.0
boxsize = 40.0, 40.0, 40.0
  • The DOCKING section includes the software that should be used for docking, and if minimization, rescoring and/or cleanup should be performed. The docking software should be specified with coma separation through the key programs. The keys relative to the DOCKING section are:
    • programs: specifies the software which are used for docking (autodock, dock6, moe and/or vina). Options relative to each program (or instance) are specfied within the section of the same name. For example, if autodock is in the list of programs, options associated with autodock should be specified in the AUTODOCK section. In case the same software needs to be used multiple times, numbering can be appended to the name of the program (e.g., in the first example below, multiple runs of MOE are performed using different scoring methods: moe, moe1, moe2).
    • minimization: performs minimization on the generated poses (yes or no).
    • rescoring: performs rescoring on the generated poses (yes or no). I strongly recommend to enable minimization in case rescoring is done. This will avoid a lot clashes, especially when the software used for rescoring are different from those used for docking. If the rescoring option is enabled, a section RESCORING should be created that contains all the options relative to that step (see below).
    • cleanup: specifies if big intermediate files should be removed (yes or no).
    • site: specifies the labels for the binding sites in case multiple binding sites are considered (site1, site2,...). See the example configuration to dock on multiple binding site, minimize and rescore the poses with multiple software.

Docking and rescoring options relative to each program are detailed in the section Docking/scoring options relative to each software

  • The SITE section includes the information about the box to spot the binding site. The keys are the following:
    • center: x, y, z coordinates of the center of the binding box (in Å).
    • boxsize: size of the box along each dimension x, y, z. The dimensions of the box should be no more than 50.0, 50.0, 50.0 (in Å).
  • The RESCORING section has only one key specifying the programs used to rescore:
    • program: specifies the software which are used for docking (autodock, dock6, moe and/or vina). Options relative to each program (or instance) are specfied within the section of the same name. For example, if autodock is in the list of programs, options associated with autodock should be specified in the AUTODOCK section. In case the same software needs to be used multiple times, numbering can be appended to the name of the program (e.g., in the example below, multiple runs of MOE are performed using different scoring methods: moe, moe1, moe2).

Each section relative to a docking/scoring program should be named the way it was specified under program in the DOCKING and/or RESCORING section. Below is a list of all the options per software that can be specified in the configuration file.

Autodock

  • ga_run (default: 100): number of autodock runs = targeted number of final poses
  • spacing (default: 0.3): grid spacing

Note 1: the partial charges of the ligand are obtained from the Gasteiger method using the AutodockTools command prepare_ligand4.py

Note 2: the number of energy evalutations ga_num_evals is automatically calculated from the number of torsions angles in the ligand structure via the formula:

ga_num_evals = min(25000000, 987500 * n_torsion_angles + 125000)

Note 3: As is usually the case for Autodock, non polar hydrogens in the ligand structure are removed prior to docking in order to properly use the Autodock force field. Once the docking has been performed, nonpolar hydrogens are reattributed in a way consistent with the input structure. Unless the minimize option in the configuration file is set to yes, no minimization is performed on those hydrogens.

Note 4 Final poses are extracted from the .dlg file using Open Babel via the following command:

obabel -ad -ipdbqt dock.dlg -omol2 lig-.mol2 -m

Autodock Vina

  • cpu (default: 1)
  • energy_range (default: 3)
  • num_modes (default: 9): targeted number of final poses

Note 1: the partial charges of the ligand are obtained from the Gasteiger method using the AutodockTools command prepare_ligand4.py

Note 2: As is usually the case for Autodock Vina, non polar hydrogens in the ligand structure are removed prior to docking in order to properly use the Autodock force field. Once the docking has been performed, nonpolar hydrogens are reattributed in a way consistent with the input structure. Unless the minimize option in the configuration file is set to yes, no minimization is performed on those hydrogens.

DOCK 6

  • attractive_exponent (default: 6)
  • extra_margin (default: 2.0)
  • grid_spacing (default: 0.3)
  • maximum_sphere_radius (default: 4.0)
  • max_orientations (default: 10000)
  • minimum_sphere_radius (default: 1.4)
  • nposes (default: 20): targeted number of final poses
  • num_scored_conformers (default 5000)
  • probe_radius (default: 1.4)
  • repulsive_exponent (default: 12)

DSX

MOE (scoring)

  • gtest (default: 0.01)
  • maxpose (default: 5)
  • placement (default: Triangle Matcher)
  • placement_maxpose (default: 250)
  • placement_nsample (default: 10)
  • remaxpose (default: 1)
  • rescoring (default: GBVI/WSA dG)
  • scoring (default: London dG)

Multi-program docking on a single binding site

Below is an example of configuration file that can be used as an input of rundbx. The docking procedure is carried out on a single binding site specied as a box with dimensions 30.0 x 30.0 x 30.0 centered at the position (x, y, z) = 8.446, 25.365, 4.394.

[DOCKING]
program = autodock, vina, dock, moe, moe1, moe2
rescoring = no
minimize = yes
cleanup = no

[AUTODOCK]
ga_run = 50
spacing = 0.3

[VINA]
num_modes = 20

[DOCK]
nposes = 200

[MOE]
scoring = London dG
maxpose = 100
remaxpose = 50

[MOE1]
scoring = GBVI/WSA dG
maxpose = 100
remaxpose = 50

[MOE2]
scoring = Affinity dG
maxpose = 100
remaxpose = 50

[SITE]
center = 8.446, 25.365, 4.394
boxsize = 30.0, 30.0, 30.0

Multi-program docking and rescoring on multiple binding sites

Below is another example of configuration file for rundbx used to dock on two binding sites and rescore with DrugScoreX (dsx), Autodock and Autodock Vina.

[DOCKING]
site = site1, site2
program = autodock, vina, dock
rescoring = yes
minimize = yes
cleanup = yes

[RESCORING]
program = dsx, autodock, vina

[DSX]
pot_dir = /pmshare/jordane/CSD_potentials/DSX_CSD_Potentials_v0511/csd_pot_0511/
other_flags = -T0 1.0 -T1 1.0 -T3 1.0 -j

[AUTODOCK]
ga_run = 20
spacing = 0.4

[VINA]
num_modes = 20

[DOCK]
nposes = 20

[SITE1]
center = 75.5, 80.0, 31.0
boxsize = 40.0, 40.0, 40.0

[SITE2]
center = 75.5, 40.0, 50.0
boxsize = 40.0, 40.0, 40.0

Note that the DOCKING section includes the label of the binding sites through the keyword site, here, site1 and site2. Each label refers to the section of the same name SITE1 and SITE2, respectively.

About

Python package to facilitate the use of popular docking software

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published