(Note: These tutorials are meant to provide
illustrative examples of how to use the AMBER software suite to carry out
simulations that can be run on a simple workstation in a reasonable period of
time. They do not necessarily provide the optimal choice of parameters or
methods for the particular application area.)
Copyright Andrew T. Fenley & Michael K. Gilson 2014
Section 3: Running the Production Simulations
Preparing the input file:
In this section, we will run the production simulations for each of the four systems. The data generated from these simulations will be used for our estimate of the binding enthalpy. The length of simulation and how often we save uncorrelated configuration energies are the critical components to generating a precise estimate of the binding enthalpy. Therefore, we will set the simulation time of each system to 500 ns and save energy information every 1 ps.
Note: the determination of simulation duration and the saving frequency is very system dependent. These host-guest systems are considerably smaller than protein-ligand systems so the duration of the simulations and time between saved snapshots used in this tutorial will almost certainly have to be increased for systems with more degrees of freedom.
The production simulations described in this tutorial are in the NVT ensemble to increase sampling speed. However, this is only recommended if one has previously determined the average box volume of the system in the NPT ensemble as discussed in the equilibration stage from Section 2. Otherwise, the use of the NPT ensemble during production runs is highly recommended since the mean absolute potential energy of a system is sensitive to the box volume associated with a target pressure. It is imperative that the simulations are all properly set to the same target pressure (usually 1 bar)!
| prod_md.in |
Production simulations &cntrl imin = 0, irest = 1, ntx = 5, ntb = 1, cut = 9.0, ntr = 0, ntc = 2, ntf = 2, tempi = 300.0, temp0 = 300.0, ntt = 3, gamma_ln = 1.0, ig = -1, iwrap = 1, ioutfm = 1, nstlim = 250000000 dt = 0.002, ntpr = 500, ntwx = 500, ntwr = -25000000 / &ewald nfft1=32, nfft2=32, nfft3=32, order=4 / |
Running the simulations
The above input file sets the simulation time to 500 ns which will take over a day and a half to complete per system using a GTX 680 GPU and pmemd.cuda. We recommend having access to multiple GPUs either locally or via a cluster so as to run all four simulations simultaneously. For local machines, remember to set the environment variable "CUDA_LOCAL_DEVICES" such that you prevent multiple simulations from running on identical GPUs.
To run the second set of NPT simulations, you can use the following commands:
pmemd.cuda -O -i prod_md.in -c water_prod.rst7 -p water.prmtop -o water_prod1.out -r water_prod1.rst7 -x water_prod1.nc
pmemd.cuda -O -i prod_md.in -c b2_prod.rst7 -p b2.prmtop -o b2_prod1.out -r b2_prod1.rst7 -x b2_prod1.nc
pmemd.cuda -O -i prod_md.in -c cb7_prod.rst7 -p cb7.prmtop -o cb7_prod1.out -r cb7_prod1.rst7 -x cb7_prod1.nc
pmemd.cuda -O -i prod_md.in -c cb7_b2_prod.rst7 -p cb7_b2.prmtop -o cb7_b2_prod1.out -r cb7_b2_prod1.rst7 -x cb7_b2_prod1.nc
The size of the NETCDF trajectory files are too large to host online. However, we created a gzip'd tarball with the output files from the production simulations: all_prod1_out.tar.gz. These files can be used in section 4 as an alternative to running the production simulations for this tutorial.
Click here to go back to section 2
(Note: These tutorials are meant to provide
illustrative examples of how to use the AMBER software suite to carry out
simulations that can be run on a simple workstation in a reasonable period of
time. They do not necessarily provide the optimal choice of parameters or
methods for the particular application area.)
Copyright Andrew T. Fenley & Michael K. Gilson 2014