Receptor with flexible side chains

Generating target file detailed explanations

The agfr command is used to compute grids of affinity values for a given set of atom types. This calculation is performed using the AutoGrid4 software program. These affinity values are calculated on evenly spaced grid points for a 3D box (docking box) placed over the receptors site targeted for automated docking.

> agfr -r 4EK3_rec.pdbqt # this is the target protein for docking

       -l 4EK4_lig.pdbqt # this is a known ligand

-P 7.0            # larger padding

-f A:ILE10,PHE82  # side chains to be made flexible

-o 4EK3_rec_FR_10_82 #save result in 4EK3_rec_FR_10_82.trg

The receptor is specified using the r/--receptor command line option. This option is required.

The position and site of the box can be specified in a variety of way using the b/--boxMode option. In this particular case we instruct agfr to create the box as the bounding box of a known ligands atom centers (-l/--ligand).

By default, a padding of 4 is added to every side of the bounding box. Since the side chains can move it is reasonable to increase the padding to ensure the side chains will remain within the box when they assume different conformations. Here, the padding is set to 7 using the P/--padding option.

The flexible side chains are specified with the f/--flexRes option. This wil prevent the C-alpha and side chain atoms from contributing to the affinity maps.

The agfr command generates a target file with a .trg extension. This file will be saved under the name specified by the o/--output command line option. When omitted, a unique and descriptive filename will be created automatically. The target file contains the calculated affinity maps, translational points for placing the ligand in sensible places in the docking box, and meta data about the gird size, position, spacing, receptor atoms involved, affinity maps, etc. These files can be inspected using the about command.

The translational points are computed using the AutoSite software program. This program will analyze affinity maps and cluster high affinity points to identify clusters of points modeling potential binding pockets. The p/--pocketMode allows specifying how to handle multiple clusters of affinity points representing the pockets found in the docking box. Since this option is omitted here, all clusters are merged to create a single set of translational points.

By default, maps are compute for all AutoDock4 atom types. The list of atom types for which to compute affinity maps can be set using the m/--mapTypes option.  Generating affinity maps for fewer atom types generates smaller target files and takes less time to perform the calculation, however, such a target file cannot be used for docking ligands containing atoms for which the target files does not contain the affinity map.

Running this command generates the following output (saved in 4EK3_rec_FR_10_82.log)

 

fiji:rib sanner$ agfr -l 4EK4_lig.pdbqt -r 4EK3_rec.pdbqt -P 7.0 -f A:ILE10,PHE82 -o 4EK3_rec_FR_10_82

MSMSLIB 1.4.4 started on fiji.scripps.edu

Copyright M.F. Sanner (March 2000)

Compilation flags 

#################################################################

# If you used AGFR in your work, please cite:                   #

#                                                               #

# P.A. Ravindranath S. Forli, D.S. Goodsell, A.J. Olson and     #

# M.F. Sanner                                                   #

# AutoDockFR: Advances in Protein-Ligand Docking with           #

# Explicitly Specified Binding Site Flexibility                 #

# PLoS Comput Biol 11(12): e1004586                             #

# DOI:10.1371/journal.pcbi.1004586                              #

#                                                               #

# <PAPER TO COME> and                                           #

#                                                               #

# P. Ananad Ravindranath and M.F. Sanner                        #

# AutoSite: an automated approach for pseudoligands prediction  #

# - From ligand binding sites identification to predicting key  #

# ligand atoms                                                  #

# Bioinformatics (2016)                                         #

# DOI:10.1093/bioinformatics/btw367                             #

#                                                               #

# Please see http://adfr.scripps.edu for more information.      #

#################################################################

 

Computing grids on fiji.scripps.edu a Darwin-14.3.0-x86_64-i386-64bit computer

Date Fri Dec 16 14:33:43 2016

 

loading receptor: 4EK3_rec.pdbqt

loading ligand: 4EK4_lig.pdbqt

 

set box using ligand

    Box center:    23.332    28.922    29.598

    Box length:    23.250    22.500    18.000

    Box size  :        62        60        48

    padding   :     7.000

    spacing   :     0.375

 

identifying pockets using AutoSite ....

    found 3 pockets

 

    pocket|  energy | # of |Rad. of | energy |   bns    | score 

    number|         |points|gyration|per vol.|buriedness|v*b^2/rg

    ------+---------+------+--------+--------+----------+---------

        1   -110.05   206    4.03     -0.53      0.85      36.76

        2    -71.00   155    3.57     -0.46      0.84      30.33

        3    -50.46    66    2.29     -0.76      0.98      27.44

    merging clusters ...

done. got 427 fill Points, in 2.21 (sec)

 

setting map types using: all to ['HS', 'Mg', 'HD', 'NA', 'Fe', 'Br', 'NS', 'A', 'C', 'Mn', 'G', 'F', 'I', 'H', 'J', 'N', 'Q', 'P', 'S', 'GA', 'Z', 'Zn', 'Cl', 'Ca', 'OA', 'SA', 'OS']

 

computing maps for center=(23.332 28.922 29.598) size=(23.250 22.500 18.000) dims=(62 60 48) ...

    422 points inside the box

 

    maps computed in 13.94 (sec)

    the following 13 flexible receptor atoms did not contribute to the grid calculation:

      A:ILE10:CA,CB,CG1,CG2,CD1,

      A:PHE82:CB,CG,CD1,CD2,CE1,CE2,CZ,CA,

    making target file 4EK3_rec_FR_10_82.trg ...    done.

done. 18.84 (sec)

 

Explanations:

After reading the receptor and ligand a box is set up around the ligand with a padding 5.0

set box using ligand

    Box center:    23.332    28.922    29.598

    Box length:    23.250    22.500    18.000

    Box size  :        62        60        48

    padding   :     7.000

    spacing   :     0.375

this box is then used to run AutoSite to identify pockets (i.e. clusters of fill points) within this box.

The resulting clusters are reported in the table

identifying pockets using AutoSite ....

    found 3 pockets

 

    pocket|  energy | # of |Rad. of | energy |   bns    | score 

    number|         |points|gyration|per vol.|buriedness|v*b^2/rg

    ------+---------+------+--------+--------+----------+---------

        1   -110.05   206    4.03     -0.53      0.85      36.76

        2    -71.00   155    3.57     -0.46      0.84      30.33

        3    -50.46    66    2.29     -0.76      0.98      27.44

    merging clusters ...

done. got 427 fill Points, in 2.21 (sec)

In this case AutoSite identified 3 clusters of affinity points corresponding to 3 pockets within the docking box. the default --pocketMode option is to use all pockets, hence, the 3 clusters of points are merged to define the translational points the adfr will use to place the ligand during docking.

The affinity maps are computed for all AutoDock4 atom types.

setting map types using: all to ['HS', 'Mg', 'HD', 'NA', 'Fe', 'Br', 'NS', 'A', 'C', 'Mn', 'G', 'F', 'I', 'H', 'J', 'N', 'Q', 'P', 'S', 'GA', 'Z', 'Zn', 'Cl', 'Ca', 'OA', 'SA', 'OS']

Next the maps and valid translational points are computed

computing maps for center=(23.332 28.922 29.598) size=(23.250 22.500 18.000) dims=(62 60 48) ...

    422 points inside the box

Note that some fill points identified by AutoSite might fall slightly outside the box. Such points are discarded reducing the number of translational points from 427 to 422.

The atoms not contributing to the grid are listed next

    the following 13 flexible receptor atoms did not contribute to the grid calculation:

      A:ILE10:CA,CB,CG1,CG2,CD1,

      A:PHE82:CB,CG,CD1,CD2,CE1,CE2,CZ,CA,

    maps computed in 13.94 (sec)

By default, AutoGrid maps are post-processed to remove translational points located inside the molecular surface and to create a potential gradient pointing form the core of the protein to the surface.

Finally, a file containing the maps, as well as the receptor and translational points (along with meta data about the maps) is created.

    making target file 4EK3_rec_FR_10_82.trg ...    done.