Amber masthead
Filler image AmberTools24 Amber24 Manuals Tutorials Force Fields Contacts History
Filler image

Useful links:

Amber Home
Download Amber
Installation
Amber Citations
GPU Support
Updates
Mailing Lists
For Educators
File Formats
Contributors
Workshops
Free Energies
 



SECTION 3: Linear interaction energy analysis


FEW provides the functionality to conduct binding free energy calculations according to the linear interaction energy (LIE) approach described in detail in [3, 4]. In this approach the binding free energy is estimated from the change in electrostatic (ele) and van der Waals (vdW) interaction energy between the ligand and its environment upon complex formation. The interaction energy contributions are determined from snapshots of MD simulations of the free and the complex bound ligand in solution by subtracting the energies of the individual components from the total energy of the solvated systems (Figure 4). Energies are determined by single point calculations with sander. The interaction energies calculated for individual snapshots are averaged over the whole conformational ensemble selected for evaluation by the user.
The binding free energy is estimated as the sum of the differences in vdW and electrostatic interaction energies scaled by weighting factors (cf. last equation in Figure 4). FEW uses established weighting factors to provide a rough estimate of the binding free energy of a ligand for the studied target system.

Attention: Despite the demonstrated capability of the employed scaling factors to give good binding free energy estimates, this might not be the case for all protein-ligand systems (see [7] for a more detailed discussion). Therefore, it is strongly recommended to thoroughly investigate whether other scaling factors are more appropriate for a specific ligand-protein system. If binding affinities have been determined experimentally for some of the studied compounds, optimal scaling factors should be determined by a linear fit of the computed energy contributions to experimentally determined binding energies.

Calculation of binding free energy by the LIE approach
Figure 4: Schematic depiction of procedure employed in FEW for the estimation of the ligand binding free energy by the LIE approach.

Prerequisite for the setup of LIE analyzes with FEW is the existence of trajectories of the ligand and the complex in solution generated using the MD setup functionality of FEW. As we have already created such trajectories in section 1 of the tutorial employing the 3-trajectory setup for MM-PBSA calculations, all necessary information is already available. As command-file for the preparation of the LIE analyzes serves the file lie_am1 located in the /home/user/tutorial/cfiles directory.

Tip: If you plan to conduct only LIE calculations, setup the MD simulations with traj=3 and use LIE instead of MMPBSA in the FEW program call. This way only MD simulations for complex and ligand are prepared and no receptor simulation, which is superfluous for LIE analyzes, is setup.

lie_am1
@LIEW
################################################################################
# Command file for LIE calculations based on trajectories
# generated by molecular dynamics simulations previously.
################################################################################
# Location and features of input and output directories / file(s)
#
# lig_struct_path: Folder containing the ligand input file(s)
# output_path: Basis directory in which all setup and analysis folders will
#              be generated. The directory needs to be identical with the
#              'output_path' directory used for setup of the MD simulations.
lig_struct_path              /home/user/tutorial/structs output_path                  /home/user/tutorial

# Receptor features
# water_in_rec: Water present in receptor PDB structure
#               used for setup of MD simulations
water_in_rec                 1

################################################################################
# General Parameters for LIE calculation setup
#
# lie_calc: Request setup of LIE calculations
# charge_method: Charge method used for MD, either "resp" or "am1"
# no_of_rec_residues: Number of residues in receptor structure
# additional_library: If an additional library file is required, e.g. for
#                     non-standard residues present in the receptor structure,
#                     this file must be specified here.
# additional_frcmod: If additional parameters are needed, e.g. for describing
#                    non-standard residues present in the receptor structure,
#                    a parameter file should be provided here.
# lie_executable: Location of LIE calculation script provided with FEW.
#                 If no script is specified, it is assumed that the LIE
#                 program is located in the default directory.
# lie_batch_template: Template batch file for LIE calculation
# lie_batch_path: Path to regard as basis path for setup of batch files
#
lie_calc                1
charge_method           am1
no_of_rec_residues      290
additional_library      
/home/user/tutorial/input_info/CA.lib
additional_frcmod
lie_executable
lie_batch_template      
/home/user/tutorial/input_info/lie.sge
lie_batch_path
#
################################################################################
# Parameters for coordinate extraction
#
# image_trajectories: If set to "1" solutes of the specified trajectories
#                     will be imaged to the origin before coordinates are
#                     extracted. Please regard that this may require a large
#                     amount of additional disc space.
# trajectory_files: Trajectory files to regard. The path will be determined
#                   automatically. Specify 'all' to regard all trajectories
#                   produced in the production phase of the MD simulations.
#                   This ensures consistent snapshot numbering.
#                   Subsets of snapshots will be generated according to the
#                   parameters first_lie_snapshot, last_lie_snapshot, and
#                   offset_lie_snapshots. If only a subset of the trajectories
#                   from MD shall be used, the individual files must be
#                   specified as 'trajectory_file ' providing one
#                   entry per line.
#                   ATTENTION: The trajectory file(s) must exist for both the
#                              simulation of the complex and the simulation of
#                              the free ligand in solution.
# snaps_per_trajectory: Number of snapshots stored in each trajectory.
#
image_trajectories        1
#
trajectory_files          md_prod_001.mdcrd.gz
trajectory_files          md_prod_002.mdcrd.gz
#
snaps_per_trajectory      50
################################################################################
# LIE Analysis
#
# calc_sasa: Request calculation of change in solvent accessible surface area
# first_lie_snapshot: No. of snapshot from which the LIE calculation
#                     shall be started
# last_lie_snapshot: No. of last snapshot that shall be regarded in the
#                    LIE calculation
# offset_lie_snapshots: Offset between snapshots that shall be regarded
#                       in the LIE calculation
# sander_executable: Sander executable that shall be used for the LIE
#                    calculation, if this differs from the default one
#                    in $AMBERHOME/bin
# parallel_lie_call: Call for running sander in parallel, e.g.:
#                    mpirun -np 2 -machinefile $PBS_NODEFILE
#                    Prerequisite: Parallel version of sander installed
# delete_lie_trajectories: Delete trajectory files created for LIE analysis
#                          directly after energy calculation
#
calc_sasa                    0
first_lie_snapshot           81
last_lie_snapshot            100
offset_lie_snapshots         1
sander_executable
parallel_lie_call            mpirun -np 8
delete_lie_trajectories      0

The FEW command-file for LIE analyzes consists of the following parts:

1. Location of input and output directories / file(s):
Here the input / output specific parameters are defined. This part of the script is identical with the corresponding part of the other command files presented before (see e.g. section 1).

2. General Parameters for LIE calculation setup:
LIE analyzes are setup based on MD simulations in which
am1 charges were used for the ligand(s). For the structurally bound calcium ion of the 290 residues comprising receptor an additional library file is loaded. The LIE.pl executable is assumed to be in the default location and LIE calculations are setup using the basic input/output directory as output directory. As template for the setup of batch files servers the file specified under lie_batch_template. In this template file the settings before the Prepare calculation section need to be adapted to the computing environment where the LIE calculations shall be performed.

3. Parameters for coordinate extraction:
This part of the command file is similar to the snapshot extraction specific part of the command file for MM-PB(GB)SA calculations described in section 2 of this tutorial.  Imaging of the trajectories before coordinate extraction is requested. All trajectories that have been produced during MD production are specified. This is required to ensure correct identification of the snapshot coordinates. The snapshots that shall be taken into account in the LIE calculations are selected based on the number of snapshots present in each trajectory file
snaps_per_trajectory as well as based on first_lie_snapshot, last_lie_snapshot, and offset_lie_snapshots in section 4 described below.

4. Parameters for LIE analyzes:
LIE analyzes are requested based on snapshots
81 to 100 with an offset of 1. FEW offers the possibility to compute the difference in the solvent accessible surface area of the ligand between the bound and the free state, since this quantity has been used to determine a ligand specific coefficient γ (cf. last equation in Figure 4) [8, 9, 10]. Here, however, this option is not selected (calc_sasa=0). A computing environment specific parallel_lie_call for sander is provided. Consequently the LIE analyzes will be run with the parallel version of sander under $AMBERHOME/bin/sander.MPI. As the trajectories that are generated for LIE analysis from the original MD trajectories contain water molecules explicitly and can therefore consume a large amount of disk space, FEW offers to delete them directly after the LIE calculation for one ligand has been completed. This option was not switched on here (
delete_lie_trajectories=0).

Tip: If you want to calculate binding free energies by LIE analysis based on a large number of snapshots, please check first the size of the disk space consumed by the files created in the setup of the calculations for a small number of snapshots and then estimate, whether there is sufficient space on your computing system.

Running FEW:
When the computing environment specific settings have been adapted and the
basic input/output path in the command file has been correctly specified, the LIE calculations can be prepared with FEW.

perl $FEW/FEW.pl LIE /home/user/tutorial/cfiles/lie_am1

The output:
After successful termination of FEW, the
/home/user/tutorial directory contains a new folder called lie_am1 with the sub-structure shown in Figure 5. Please note, the energy calculation folders depicted in violet in Figure 5 are not present yet. They will be created during the LIE calculations.


Sub-structure of the folder created in setup of LIE analyzes
Figure 5: Schematic depiction of sub-structure of the folder created upon LIE analysis using FEW.

Starting the LIE calculations:
LIE calculations can be started by executing the shell script
qsub_LIE.sh located in the lie_am1 folder.

/home/user/tutorial/lie_am1/qsub_LIE.sh


LIE results:
The final results of the LIE calculations can be found in the ligand specific sub-folders of the
/home/user/tutorial/lie_am1 directory. The file LIE_s<S>_<T>_<O>.txt with <S> = 
first_lie_snapshot, <T> = last_lie_snapshot, and <O> = offset_lie_snapshots, i.e. in case of this tutorial the file LIE_s81_100_1.txt, lists the calculated average electrostatic and vdW interaction energies (and the solvent accessible surface area) for the complex bound and the free ligand, as well as their differences. Furthermore an estimate of the binding free energy using the scaling factors shown in Figure 4 is given.

Result file LIE_s81_100_1.txt of ligand L51c
Energy contributions:
                           Mean energy    Standard deviation    Standard error
Complex non-bonded VDW:     -76.550390              3.354497          0.750088
Complex non-bonded ELE:     -49.058070              7.613551          1.702442
Ligand non-bonded VDW:      -43.399665              3.762162          0.841245
Ligand non-bonded ELE:      -56.828235              6.755975          1.510682


================================================================================
Final LIE results:

Differences in interaction energies and SASA:

Delta electrostatic energy (dE-ele):      7.770165 kcal/mol
Delta van der Waals energy (dE-vdW):    -33.150725 kcal/mol
________________________________________________________________________________

Estimated binding free energy:

ATTENTION: Please note that the estimate of the binding free energy (dG-bind)
           provided below was calculated according to the equation
           dG-bind = alpha * dE-vdW + beta * dE-ele
           with fixed coefficients of alpha=0.16 and beta=0.5.
           We strongly recommend to thoroughly investigate whether using other
           coefficients could be beneficial in the specific case.

Binding energy estimate (dG-bind):       -1.419034 kcal/mol

Tip: If you wish to determine by a linear fit the best weighting coefficients for a data set for which experimental binding affinities are known, you can extract the differences in electrostatic and vdW interaction energies (as well as the solvent accessible surface area) from all result files of a dataset with the script extract_LIEenergies.pl provided with FEW. Please refer to the FEW manual for details.


SECTION 2: MM-PB(GB)SA calculations

SECTION 4: TI calculations

Copyright: Nadine Homeyer 2014