fftdock (c46b2)
FFTDOCK: An Accelerated Routine for Rigid-Protein Rigid-Ligand Docking
Using Fast Fourier Transform Algorithm
By Xinqiang Ding and Charles L. Brooks III
* Background: Background information about the method
* Syntax: Syntax of the command
* Description: Description of the command
* Examples: Example commands
* Installation: How to compile fftdock
Using Fast Fourier Transform Algorithm
By Xinqiang Ding and Charles L. Brooks III
* Background: Background information about the method
* Syntax: Syntax of the command
* Description: Description of the command
* Examples: Example commands
* Installation: How to compile fftdock
Top
Background Information
Fast Fourier transform (FFT) algorithm has been successfully used for rigid
protein-protein docking. The basic idea is as the following. The interaction
energy between two proteins (in fixed orientations) for all possible relative
positions can be formulated as a cross correlation function between two 3
dimenstional grid, each of which can be calculated based of a protein's
structure. The FFT algorithm can be used to greatly accelerate the calculation
of the cross correlation function. Here the same idea is adapted for protein-
ligand docking. Given a protein structure, potential grids can be calculated
around the ligand binding pocket (» grid for further information). Given
a ligand with a fixed conformation and a fixed orientation, similar grid be
calculated. Then the FFT algorithm is used to calculate interaction energies
between the proteina and ligand for all possible translations of the ligand
inside the binding pocket. The FFT calculation for multiple ligand orientations
are parallized on graphic processing units (GPUs).
Background Information
Fast Fourier transform (FFT) algorithm has been successfully used for rigid
protein-protein docking. The basic idea is as the following. The interaction
energy between two proteins (in fixed orientations) for all possible relative
positions can be formulated as a cross correlation function between two 3
dimenstional grid, each of which can be calculated based of a protein's
structure. The FFT algorithm can be used to greatly accelerate the calculation
of the cross correlation function. Here the same idea is adapted for protein-
ligand docking. Given a protein structure, potential grids can be calculated
around the ligand binding pocket (» grid for further information). Given
a ligand with a fixed conformation and a fixed orientation, similar grid be
calculated. Then the FFT algorithm is used to calculate interaction energies
between the proteina and ligand for all possible translations of the ligand
inside the binding pocket. The FFT calculation for multiple ligand orientations
are parallized on graphic processing units (GPUs).
Top
Syntax of Commands
Initialize Grid Potential:
FFTG READ UNIT <integer> [Formatted]
Run Docking:
FFTG LCON -
NCON <integer> ICON <integer> -
NROK <integer> SIZB <integer> -
QUAU <integer> NQUA <integer> -
IQUA -
SELE <atom selection> END
Collect Results:
FFTG COOR -
ICON <integer> IROT <integer>
Syntax of Commands
Initialize Grid Potential:
FFTG READ UNIT <integer> [Formatted]
Run Docking:
FFTG LCON -
NCON <integer> ICON <integer> -
NROK <integer> SIZB <integer> -
QUAU <integer> NQUA <integer> -
IQUA -
SELE <atom selection> END
Collect Results:
FFTG COOR -
ICON <integer> IROT <integer>
Top
Detailed Description of Options in Commands
Initialize Grid Potential:
FFTG READ UNIT <integer> [Formatted]
UNIT Unit from which to read the protein grid potentials
Formatted Set if the grid potential is in ASCII format
This command reads in protein grid potentials generated using grid command (see
doc/grid.doc).
Run Docking:
FFTG LCON -
NCON <integer> ICON <integer> -
NROK <integer> [SIZB <integer>] -
QUAU <integer> NQUA <integer> -
[IQUA] -
SELE <atom selection> END
NCON Num of ligand conformations which will be docked
ICON Index of current ligand conformation
NROK Num of low energy ligand orienations saved
SIZB Size of batch for FFT
QUAU Unit num of file from which quaternions are read
NQUA Num of ligand orientations for each conformation
IQUA Set if the original orientation is included
This command does the FFT-based docking. A ligand is specified using an atom
selection. Its conformation coordinates are read from main coordinates. For
each conformation, NQUA uniformly distributed orientations are generated and
docked using FFTs. The FFTs are run in batches with a batch size of SIZB. It
is expected that a large batch size should be used as along as it can fit into
the GPU device's memory, because a large batch size can make better use of the
parallelism provide by GPUs. After the FFT dock, these orientations are ranked
based on its energy and NROK low energy orientations are saved internal.
Collect Results:
FFTG COOR -
ICON <integer> IROT <integer>
ICON Index of conformation
IROT Index of orientation
This command copies the ligand coordinates corresponding to ICON'th
conformation and IROT'th orientaion to the main coordinates.
Detailed Description of Options in Commands
Initialize Grid Potential:
FFTG READ UNIT <integer> [Formatted]
UNIT Unit from which to read the protein grid potentials
Formatted Set if the grid potential is in ASCII format
This command reads in protein grid potentials generated using grid command (see
doc/grid.doc).
Run Docking:
FFTG LCON -
NCON <integer> ICON <integer> -
NROK <integer> [SIZB <integer>] -
QUAU <integer> NQUA <integer> -
[IQUA] -
SELE <atom selection> END
NCON Num of ligand conformations which will be docked
ICON Index of current ligand conformation
NROK Num of low energy ligand orienations saved
SIZB Size of batch for FFT
QUAU Unit num of file from which quaternions are read
NQUA Num of ligand orientations for each conformation
IQUA Set if the original orientation is included
This command does the FFT-based docking. A ligand is specified using an atom
selection. Its conformation coordinates are read from main coordinates. For
each conformation, NQUA uniformly distributed orientations are generated and
docked using FFTs. The FFTs are run in batches with a batch size of SIZB. It
is expected that a large batch size should be used as along as it can fit into
the GPU device's memory, because a large batch size can make better use of the
parallelism provide by GPUs. After the FFT dock, these orientations are ranked
based on its energy and NROK low energy orientations are saved internal.
Collect Results:
FFTG COOR -
ICON <integer> IROT <integer>
ICON Index of conformation
IROT Index of orientation
This command copies the ligand coordinates corresponding to ICON'th
conformation and IROT'th orientaion to the main coordinates.
Top
! read grid potential
open unit 30 read unform name "grid.bin"
FFTG READ UNIT 30
! read different conformations and dock them
FFTG LCON NCON 10 ICON 1 NROK 2 NQUA 100 IQUA sele segi LIGA end
...
add more commands to read in another ligand conformation
...
FFTG LCON NCON 10 ICON 2 NROK 2 NQUA 100 IQUA sele segi LIGA end
! save ligand structures after docking
FFTG COOR ICON 1 IROT 1
...
add more commands to save the current structure
...
FFTG COOR ICON 1 IROT 2
! apparently you can also write loop to go through all the docked structures.
! read grid potential
open unit 30 read unform name "grid.bin"
FFTG READ UNIT 30
! read different conformations and dock them
FFTG LCON NCON 10 ICON 1 NROK 2 NQUA 100 IQUA sele segi LIGA end
...
add more commands to read in another ligand conformation
...
FFTG LCON NCON 10 ICON 2 NROK 2 NQUA 100 IQUA sele segi LIGA end
! save ligand structures after docking
FFTG COOR ICON 1 IROT 1
...
add more commands to save the current structure
...
FFTG COOR ICON 1 IROT 2
! apparently you can also write loop to go through all the docked structures.
Top
To install CHARMM with the fftdock command, give the configure script the
--with-fftdock
command line option. For example:
configure --with-fftdock ...
To install CHARMM with the fftdock command, give the configure script the
--with-fftdock
command line option. For example:
configure --with-fftdock ...