Hello,
Apologies in advance for a lengthy response! I tried to answer as
completely as I could, there's a tldr below.
In each of the gist output data explorer files (.dx) will be a spatially
resolved statistical evaluation of solvent properties within the gridded
region defined by the gist command used to run gist in cpptraj. This means
that in each voxel of the grid the entropy of water molecules found in that
voxel are being determined, statistically quantified, and then mapped into
the .dx output in that voxel, for all of the frames that are provided when
running the cpptraj command.
Convergence is an evaluation of whether the quantities being assessed are
stable over some stretch of time or number of samples (frames in MD). To
assess the convergence of gist quantities you will need to perform gist on
different temporal slices of your MD trajectories. Also it is most sensible
to assess whether the quantities of the solvent are converged with regards
to some region of interest (say a binding site of a protein/around a ligand
of interest as you mentioned, or the first solvation shell around a solute
as examples)...you could perform an analysis of the water in the full
simulation box, but most likely you are investigating a specific region, so
I would recommend performing these analyses on that region as that will be
most pertinent to your study.
To do this you will need to run gist on several segments of your MD
simulation, for instance in 10 equal segments (ie first 10th, second
10th,etc). This will require 10 separate gist calculations wherein each you
provide different sets of frames from your MD simulation (or however many
segments seems sensible for your simulations). The plots in the 2016 gist
paper are plotting out the quantities for such segments of the simulations
in that study, so you can see that at the beginning of the simulation the
quantities are quite different from what they are at the end, but at some
point the quantities converge to some final value and remain at that value.
Typically, if a quantity remains relatively stable (within some tolerance)
we consider it to be converged. In that paper we had calculated convergence
for a binding site region, which we defined using gistpp to flag voxels
within 3 Å of the cocrystal ligand pose. The same approach could be used to
define a region of interest of any size. You do not need to use a ligand,
any pdb file will work with the same gistpp commands to define a region of
interest.
So to summarize more briefly (TLDR):
To evaluate convergence first calculate gist quantities on varying segments
of your simulation (different slices of frames), then sum the gist
quantities of interest over the region of interest in those frame slices
(gistpp, page 4 of the gist tutorial can help here), plot out the summed
quantities for these simulation segments to obtain similar plots to the
2016 paper. The more slices you take the better your temporal resolution of
this analysis will be.
On a side note: for the most part this analysis is reporting on how
converged the properties of water in the region of interest are, though
gist is the tool performing that characterization, the quantities are being
directly derived from the sampled simulation frames. Therefore if at the
end of this analysis you find that the quantities are not converged the
first culprit to investigate is whether your simulation has adequately
converged in your region of interest (ie binding site).
Hope this helps!
Best,
--Steve
On Sun, Oct 15, 2023 at 11:05 AM Shakuntala Dhurua via AMBER <
amber.ambermd.org> wrote:
> Dear user,
>
> I am trying to calculate solvent entropy (for water) by using GIST present
> around the ligand molecule. I have tried tutorial given by AMBER(
> http://ambermd.org/tutorials/advanced/tutorial25/). I got many output
> file in .dx format. I have also got these two entropy output file =
> gist-dTSorient-dens.dx, gist-dTStrans-dens.dx. I am unable to understand
> these two file to interpret the convergence of entropy data. I am able to
> visualise these .dx format file by using vmd.
>
> Can someone tell me how to calculate the convergence of entropy from
> output of GIST entropy data.
>
> I have attached output file I got.
>
> I have also attached a paper where convergence of entropy was plotted
> against time.
>
> Any help from you would be highly appreciated.
>
>
> Thank you in advance.
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Oct 18 2023 - 11:00:03 PDT
