Binding Response (BR)

CCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCC
C                                Citation                                      C
C     Shijun Zhong and Alexander D. MacKerell,                                 C
C     Binding Response: A Descriptor for Selecting Ligand Binding Site         C
C     on Protein Surfaces. J. Chem. Inf. Model. 2007, 47(6), 2303-2315.        C
CCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCC

I.   INTRODUCTION
II.  DOWNLOAD
III. INSTALLATION
IV.  COMPUTATION
     1. The primary level
            Input
            Output 
     2. The extended level 
            Input
            Output
     3. User's computation


I. INTRODUCTION

Binding Response (BR) is a method to select ligand binding sites on protein surfaces for virtual database screening. The binding response values are designed to be in the range [0, 1]. A good binding site should receive a relatively large binding response value. See original manuscript for more details.

Evaluation of the binding response value requires calculation of the binding energy and the geometrical pose for a set of probe ligands docked against the concerned binding site. Therefore, a test docking of a set of small molecules needs to be performed in advance. Currently, the software DOCK is used to generate the binding poses and energy scores which are recorded in the *.mol2 format recongnizable to the present program.

The current package provides three options to perform the calculations. The first, partial option is the calculation of the binding response values for a binding site based on previously generated docking results. The second, extended option is a complete procedure for identifying binding sites starting from generating the protein surface and expressing the binding pockets, through docking, and ending with the ranking of binding sites. The remainder of this document describes the implementation of options 1 and 2. The third option is an alternate collection of scripts to perform the complete binding response calculations by Kenno Vanommeslaeghe; see bindres.html in the bindres subdirectory.


II. DOWNLOAD

Please email alex@outerbanks.umaryland.edu and request the package "binding_response.tar.gz". Use the following command to extract this gzipped file.

  tar -xvzf binding_response.tar.gz
A directory "binding_response" will be automatically generated. It contains the following files and directories.

  README.html    this file
  bin            directory of executable files: br.exe and some scripts
  src            directory of the source code
  examples       directory of examples for running br.exe
  demo           directory for demo calculation for the second, extended option
  parameter      directory of parameter files and input files
  test           work directory for performing the individual user's calculations
  user_modified  directory for future user scripts
  bindres        directory containing alternate scripts for calculation of the binding response


III. INSTALLATION

All executable files in the package were tested in a Linux environment and are ready for running the first, partial option. The source code is included in the src directory.

The calculation for the second, extended options requires a series of calculations using other programs, i.e. CHARMM, DMS, and DOCK. These programs should be installed in advance to generate the following executable files:

 
  charmm, dms, sphgen, showbox, grid, dock.  
Copy these files to the directory "binding_response/bin" so that it does not need to change the path defined in the script file "binding_response/bin/br.csh".

No parameter file is required for the CHARMM calculation. However, the docking calculation needs the DOCK parameter files:

  chem.defn, flex.defn, flex_drive.tbl, vdw_AMBER_parm99.defn
Copy them to the subdirectory "binding_response/parameter" so that it is not necessary to change the path defined in the DOCK4.0 input files "grid.in" and "dock.in" which can be modified to fit user's DOCK version.


IV. COMPUTATION

1. First option

The following example describes how to compute the binding response value for a binding site.

 cd binding_response/examples
 ../bin/br.exe < ex1.inp > ex1.out

Input

The input file "ex1.inp" requires only two file names in separate lines, i.e.

       ex1.mol2
       ex1.sph
ex1.mol2 is a previously generated DOCK mol2 file which contains the energy score and the coordinates of the ligands previously docked to the binding site. It is suggested that the summation of electrostatic and van del Waals interactions be used as the energy score in the docking. Currently, the mol2 files generated by DOCK4, DOCK5, and DOCK6 can be recognized by the program.

ex1.sph, in the format of the SPHGEN output, contains a set of spheres which defines the effective space of the binding pocket. It should be pointed out that the sphere set can be different from that used in docking.

Output

The output gives the averaged binding response (BR) values for the first 10 compounds, for the first 100 compounds, ..., and for all compounds.

     Ligand_N    E_aver      BR
         10    -34.4760      0.9985
        100    -28.7006      0.9729
        108    -27.9581      0.9663

For example, the Binding Response value is 0.9663 for the binding site defined by the sphere set (ex1.sph) for the 108 docked compounds (ex1.mol2).



2. Second, extended optin

The following demo calculation shows the entire procedure of ranking binding sites on protein surfaces.

  cd binding_response/demo
  ./demo.csh &

Input

The executable file "demo.csh" will automatically read two protein structure files

  protein.pdb
  protein.mol2
which exist in the current directory.

"protein.pdb" contains no hydrogen atoms and will be called by the program "SPHGEN" to generate the protein surface.

"protein.mol2" will be used in docking and therefore contains hydrogen atoms and atomic partial charges.

Output

The output file "br.out" will include:

The top 10 binding sites evaluated based on the binding energy and pose of
2 ligands docked against each of these binding sites.

Preferred binding sites will have a larger BR value 
as given in the last column of the following table.

Binding_Site    Spheres  E_averaged BR(all_ligands)
pdbid_clust_6        16    -27.3750          0.9672
pdbid_clust_9        16    -24.7800          0.9298
pdbid_clust_8        22    -20.5750          0.9292
pdbid_clust_2        26    -20.3750          0.9288
pdbid_clust_5        16    -18.7650          0.9120
pdbid_clust_1        25    -23.8150          0.8959
pdbid_clust_7        16    -19.5150          0.8686
pdbid_clust_4        15    -20.1800          0.7701
pdbid_clust_3        17    -20.5150          0.7621
pdbid_clust_10       14    -16.4300          0.5309

Thus, the binding site labelled "pdbid_clust_6" is highest ranked binding site with the largeest Binding Response value: 0.9672.



3. User's computation

For application of the method on alternate proteins

 cd binding_response/test
Copy your protein pdb and mol2 files to the current directory and rename them to:
  protein.pdb
  protein.mol2
Type the command:
  ../bin/br.csh &
It takes about 60 hours of CPU time on an 2GHz 64bit AMD Athlon with 1GB of memory to complete the full calculation, including all docking. Then check the output file "br.out" for the results.