Hello,
Here, the cv_min and cv_max are only to set the range of output biasing
potential, not the range of
sampling. To restrain the CV in the sample, you should add another &pmd
module. like
For Amber16, if you are using GPU, you should apply this patch first and
then you can use &pmd in GPU
http://ambermd.org/tutorials/advanced/tutorial31/nfe_advance.patch.tar.gz
There is README file and also you can check the new Amber 2017 manual for
the usage of &pmd.
For wt_bias_1.nc and wt_bias_2.nc, they are the same, so choosing either of
them is OK.
For the selection_constant, this is the original paper of this idea
http://pubs.acs.org/doi/abs/10.1021/ct900524t
Here the constant is from the equation and is associated with the selection
time period and
density of CV. I also check the paper and it is hard to find a practical
way to choose the constant.
Actually the selection algorithm is to enhanced the sampling and not always
necessary to apply,
often if the sampling rate is not so slow you can leave out the selection
algorithm.
Here two replicas are too few for selection, if you want to use selection I
suggest at least 4 replicas
should be used.
Best
Feng Pan
On Tue, Jun 13, 2017 at 4:55 PM, FabrÃcio Bracht <fabracht1.gmail.com>
wrote:
> Dear All,
> I've simulated a protein in order to back up some experimental data
> regarding the importance of a Tyr residue. This residue is part of a loop
> that has an open/close type motion. The study began because the difraction
> structure showed that this Tyr is quite different from other structures
> known in the literature. With MD we have been able to see that this Tyr
> residue does something close to a pendulum motion. It prefers some
> configurations over others, meaning, it prefers either to be facing the
> catalytic center or it prefers to face the opposite way. This hinge motion
> is associated with changes in the loop conformation and it seems that the
> outward/inward positions are related to the phi/psi dihedral angles of this
> particular residue (or a neighboring residue, we are still not sure). I've
> worked with umbrella sampling before in amber, but I would like to use the
> abmd suite to calculate the relative free energy differences between these
> two configurations.
> I've started following TUTORIAL A31, specifically part 1 (Calculating 2-d
> free energy landscapes for a di-alanine peptide), but I am having some
> trouble figuring out some things.
> First, as a test run, I've adapted the files from the tutorial in order to
> describe the two TORSIONS I would like to sample. The mdin abmd variables
> are set as:
>
> &abmd
>
> mode = 'FLOODING'
>
> timescale = 1.0
> monitor_freq = 500
> monitor_file = 'abmd_monitor_1.dat'
> cv_file = 'cv.in'
> umbrella_file = 'bias_1.nc'
>
> selection_freq = 5000
> selection_constant = 0.00001
> selection_epsilon = 0.0
> wt_temperature = 10000
> wt_umbrella_file = 'wt_bias_1.nc'
>
> And the collective variable file 'cv.in' is:
>
> &colvar
> cv_type = 'TORSION'
> cv_ni = 4
> cv_i = 4270,4272,4274,4291 !phi
> resolution = 0.2
> cv_min = -1.40
> cv_max = 1.40
> /
>
> &colvar
> cv_type = 'TORSION'
> cv_ni = 4
> cv_i = 4291,4272,4274,4303 !psi
> resolution = 0.2
> cv_min = -0.7
> cv_max = 0.7
> /
>
> There are some things I would like to ask though. First, I've used the
> cv_min and cv_max to specify that I would only like to sample torsion
> values that ranges from the maximum and minimum values I observed in the MD
> simulation. Is that necessary, or even good practice?
> For this test, I am running only 2 replicas. From the tutorial, I could see
> that the wt_bias_#.nc files are equal (I've used cmp wt_bias_1.nc
> wt_bias_2.nc). So, it would not matter which file I choose for post
> processing with nfe-umbrella-slice, correct?
> The manual states that choosing the best selection_constant is somewhat of
> an art form. Can you suggest some literature (or point it out explicitly)
> as to how I would choose the best values for my simulations?
>
> Thank you
> FabrÃcio Bracht
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
--
Feng Pan
Ph.D. Candidate
North Carolina State University
Department of Physics
Email: fpan3.ncsu.edu
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Tue Jun 13 2017 - 15:00:02 PDT
