Hello Yulian,
I think it would be really helpful if we knew exactly how you are
determining dE, dS, etc. In particular how the volume around the
configurations is defined.
My first guess would be that if the total number of voxels being compared
between rigid and flexible is the same then said volume may not encompass
3.5 A around all flexible configurations (which might explain an unexpected
difference in dE).
If the system is not rotating and translating it's possible to compare
solvation by GIST without clustering, but it might require cleverly
defining the box/volume of integration. Always happy to help,
--Steve
On Wed, Nov 11, 2015 at 7:21 AM, Yulian Gavrilov <zzeppelin87.gmail.com>
wrote:
> Dear Steven and Jonathan,
>
> Thank you for your suggestions.
> Please note that N and C terminals of the flexible peptide are fixed in the
> space, so there are no rotations/translations of the whole peptide.
> When I analyze the first hydration shell (3.5A) around my rigid and
> flexible peptides I noticed strange results for dE.
> In principle, for the flexible peptide dE of the hydration water should be
> less favourable than for the frozen one (opposite to dS).
> I see it for the average value [kcal/mol per non-zero voxel] but in case of
> summation [kcal/mol for all hydration shell] dE is more favourable in case
> of the frozen peptide, which is strange.
> I expected that this may be because of some 'bad' voxels but I didn't find
> 'very bad' voxels in the flexible systems (with energies > 0 kcal/mol).
> Probably it is related to the fact that the number of non-zero voxels
> within the hydration shell is ~twice bigger for the flexible peptide (the
> total number of voxels is identical).
> Interesting, that for dS everything is fine (dS of water around the
> flexible peptide is more favourable).
> It seems that in order to make a correct comparison of the flexible and
> rigid peptides I need to use something like clustering.
>
>
>
>
> On Tue, Nov 10, 2015 at 5:28 PM, Jonathan Gough <
> jonathan.d.gough.gmail.com>
> wrote:
>
> > Hi Yulian Gavrilov,
> >
> > What Steve said is spot on. In the present implementation of
> gist-cpptraj,
> > the energy calculations utilize the periodic box dimensions (I believe
> it's
> > from the first frame) when calculating energy. When one uses the cpptraj
> > rms function the entire box is rotated at every frame. Therefore the
> change
> > in box position is lost. As a result the energy calculations are all
> wrong.
> >
> > How were you planning on calculating the desired values(dA, dTS, dE)?
> >
> > If you can simply calculate the values by integrating over the entire box
> > volume, you could run the simulation restrained and unrestrained. Then,
> > make your gist box large enough (probably best to do the same/identical
> > size about both simulations) to encompass the peptide and run gist. Then
> > you could utilize the post processing tool, gistpp, to get the values you
> > wish for each system and then compare.
> >
> > you can find the instructions for gistpp on the last page of the
> tutorial:
> >
> > http://ambermd.org/tutorials/advanced/tutorial25/section4.htm
> >
> >
> > You can grab the code from the above page or from github.
> >
> > https://github.com/gosldorf/gist-post-processing
> >
> > Don't hesitate to reach out if you need additional help.
> >
> > Best,
> > Jonathan
> >
> >
> > On Tue, Nov 10, 2015 at 10:04 AM, Steven Ramsey <vpsramsey.gmail.com>
> > wrote:
> >
> > > You can certainly analyze solvent properties around a flexible system,
> > > however there are some things to keep in mind:
> > >
> > > The GIST grid itself will not move with your system (neither will each
> > > individual voxel), therefore the sampling in certain voxels,
> particularly
> > > those that are near your solute, will have lower sampling due to solute
> > > fluctuations, which may effect several quantities (g(O), S, E).
> > >
> > > You could align the flexible system (or cluster it as you suggested),
> but
> > > then this may introduce another potential issue which was brought up in
> > > this thread:
> > >
> > > http://archive.ambermd.org/201508/0149.html
> > >
> > > Entropy should be fine on a clustered system, provided there's enough
> > > frames to sample from. The lesser populated clusters may produce
> > > thermodynamics that are not-trustworthy (meaning not converged, but
> that
> > > applies to all quantities not just S).
> > >
> > > So to summarize a bit, it is definitely possible to analyze a flexible
> > > system, but it is more troublesome than a rigid one. A lot depends on
> > just
> > > how flexible it is. Either choice (running GIST on a flexible
> simulation
> > or
> > > a clustered one) will map solvation data to the defined voxels, so you
> > can
> > > potentially play around with it and see what works for your system.
> > >
> > > Hope this helps,
> > >
> > > --Steven Ramsey
> > >
> > >
> > > On Tue, Nov 10, 2015 at 3:24 AM, Yulian Gavrilov <
> zzeppelin87.gmail.com>
> > > wrote:
> > >
> > > > Dear all,
> > > >
> > > > I have a question about GIST.
> > > >
> > > > In the tutorial and manual it is written that the protein should be
> > rigid
> > > > in order to use GIST to analyze water around it.
> > > >
> > > > Is there any way to analyze hydration water around a flexible
> molecule?
> > > > I would like to compare the thermodynamic parameters (dA, dTS, dE) of
> > > > hydration water around the flexible and rigid version of a peptide.
> Is
> > > > there any way to do it?
> > > > I use AmberTools15.
> > > >
> > > > Only idea that I have is to devide the output trajectory of the
> > flexible
> > > > peptide into several clusters based on RMSD, so that within the
> cluster
> > > the
> > > > flexibility of a peptide will be minimal. Then I can use GIST for
> each
> > > > cluster separately. But this way I won't get the right entropy
> values,
> > > > right?
> > > >
> > > > Thank you in advance for any ideas!
> > > >
> > > >
> > > > --
> > > >
> > > > Sincerely,
> > > >
> > > > Yulian Gavrilov
> > > > _______________________________________________
> > > > 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
> > >
> > _______________________________________________
> > AMBER mailing list
> > AMBER.ambermd.org
> > http://lists.ambermd.org/mailman/listinfo/amber
> >
>
>
>
> --
>
> Sincerely,
>
> Yulian Gavrilov
> _______________________________________________
> 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 Nov 11 2015 - 09:00:03 PST
