Hi Steve,
Helps a lot, thanks for clarifying!
Best,
Sérgio
************************************************
Sérgio Marques, PhD
Marie Curie Fellow at:
*Loschmidt Laboratories*
Department of Experimental Biology
Faculty of Science, Masaryk University
Kamenice 5/A13, 625 00 Brno, Czech Republic
Web: http://smarques.weebly.com/
On 24 October 2016 at 19:13, Steven Ramsey <vpsramsey.gmail.com> wrote:
> Hi Sergio,
>
> That's a nice way to use gistpp to eliminate high energy results.
>
> Unfortunately removing these high energies from your analysis will not fix
> the issue, all of the energetic values produced by GIST are performed using
> a pairwise distance, when the system is realigned these distances are
> compromised. Despite the other energies appearing to be normal it is hard
> to have confidence in their accuracy. The best way to run GIST currently is
> to use a restrained system as Jonathan suggested in the previous post. The
> entropies you are calculating should work in a re-aligned system.
>
> Hope this helps,
>
> --Steve
>
> On Mon, Oct 24, 2016 at 3:22 AM, Sérgio Marques <smar96.gmail.com> wrote:
>
> > Hi Jonathan,
> >
> > I was thinking that maybe we might use gistpp to set up some cutoff for
> the
> > energies per voxel, like -10 < Etot < 10 kcal/mol, so we could discard
> > those points with obviously wrong values (e.g.: ./gistpp -i gist-Etot.dx
> > -i2 gist-Etot.dx -op clear2 -opt cutoff1 -10.0 -opt gt1 -opt cutoff2 10
> > -opt lt2 -o correctedEn.dx). Visualizing those few high energies regions,
> > they always locate near the edges of the box, so in reality those points
> > should not be affected by the residue that I'm focused on (in the center
> of
> > the box).
> > Do you think this wouldn't be enough to correct the miscalculations?
> >
> >
> >
> > ************************************************
> > Sérgio Marques, PhD
> >
> > Marie Curie Fellow at:
> > *Loschmidt Laboratories*
> > Department of Experimental Biology
> > Faculty of Science, Masaryk University
> > Kamenice 5/A13, 625 00 Brno, Czech Republic
> > Web: http://smarques.weebly.com/
> >
> > On 21 October 2016 at 20:17, Jonathan Gough <jonathan.d.gough.gmail.com>
> > wrote:
> >
> > > Unfortunately, That explains your weird results. You can trust the
> > density
> > > and maybe a some of the other things, but the energy is going to be
> > > significantly off.
> > >
> > >
> > > One alternative is to look at the rmsf of the backbone and look for
> atoms
> > > that don't move. You could then re-run the MD keeping a few points
> fixed
> > > with loose restraints. That would keep the system in one place, then
> run
> > > gist. there will be no need for alignment as it will be stuck in one
> > > position.
> > >
> > > Hope that helps.
> > >
> > > On Fri, Oct 21, 2016 at 7:13 AM, Sérgio Marques <smar96.gmail.com>
> > wrote:
> > >
> > > > Hi Jonathan,
> > > >
> > > > My system has no restraints. I always center, align and image my
> > systems
> > > > before analyzing them. For this one in particular, I also removed the
> > > ions.
> > > >
> > > > Regards,
> > > > Sergio
> > > >
> > > > ************************************************
> > > > Sérgio Marques, PhD
> > > >
> > > > Marie Curie Fellow at:
> > > > *Loschmidt Laboratories*
> > > > Department of Experimental Biology
> > > > Faculty of Science, Masaryk University
> > > > Kamenice 5/A13, 625 00 Brno, Czech Republic
> > > > Web: http://smarques.weebly.com/
> > > >
> > > > On 20 October 2016 at 17:51, Jonathan Gough <
> > jonathan.d.gough.gmail.com>
> > > > wrote:
> > > >
> > > > > Sérgio,
> > > > >
> > > > > A few minor questions:
> > > > > 1. is your system restrained at all?
> > > > > 2. after the MD are you doing any post-processing to your
> trajectory
> > > > BEFORE
> > > > > doing gist:
> > > > > -- Are you aligning your trajectory to a structure or
> residues
> > > > (RMS)?
> > > > > -- Are you imaging your trajectory?
> > > > >
> > > > > Thanks,
> > > > > Jonathan
> > > > >
> > > > > On Thu, Oct 20, 2016 at 11:38 AM, Sérgio Marques <smar96.gmail.com
> >
> > > > wrote:
> > > > >
> > > > > > Hi Dan,
> > > > > >
> > > > > > What you say makes sense! But in my case the edges of the gist
> box
> > > are
> > > > > > still far from the PBC border. However, the truth is that those
> > very
> > > > high
> > > > > > values of energy are always near the edges of the box...
> > > > > >
> > > > > > But thanks for your comment!
> > > > > >
> > > > > > ************************************************
> > > > > > Sérgio Marques, PhD
> > > > > >
> > > > > > Marie Curie Fellow at:
> > > > > > *Loschmidt Laboratories*
> > > > > > Department of Experimental Biology
> > > > > > Faculty of Science, Masaryk University
> > > > > > Kamenice 5/A13, 625 00 Brno, Czech Republic
> > > > > > Web: http://smarques.weebly.com/
> > > > > >
> > > > > > On 20 October 2016 at 15:33, Daniel Roe <daniel.r.roe.gmail.com>
> > > > wrote:
> > > > > >
> > > > > > > Hi,
> > > > > > >
> > > > > > > On Wed, Oct 19, 2016 at 4:56 PM, Steven Ramsey <
> > > vpsramsey.gmail.com>
> > > > > > > wrote:
> > > > > > > >
> > > > > > > > There is a known issue when applying GIST to unrestrained
> > systems
> > > > > that
> > > > > > > are
> > > > > > > > then aligned using rms in cpptraj that produces arbitrarily
> > high
> > > > > > energies
> > > > > > > > and that sounds to be the issue at hand with your Eww being
> > > > 1.12E+06.
> > > > > > If
> > > > > > > > possible you may want to avoid aligning the system prior to
> > > running
> > > > > > GIST.
> > > > > > >
> > > > > > > Just to expand on this a bit, this is an unavoidable
> consequence
> > of
> > > > > > > using periodic boundary conditions. By default, GIST will use
> the
> > > > > > > minimum imaged distance between particles when performing the
> > > > > > > non-bonded energy calculation for trajectories containing unit
> > cell
> > > > > > > information. If you perform a system alignment prior to GIST
> (or
> > > > > > > really any analysis that performs imaging), the system will no
> > > longer
> > > > > > > be aligned with the original unit cell, which usually leads to
> > > > clashes
> > > > > > > between particles and imaged particles and hence the high
> > energies.
> > > > > > > You can easily illustrate this in 2 dimensions with two pieces
> of
> > > > > > > paper (the paper being your unit cell). You can lay the papers
> > side
> > > > by
> > > > > > > side so that the edges align perfectly. However, if you then
> > rotate
> > > > > > > both pieces of paper by 45 degrees clockwise the papers will
> > > overlap,
> > > > > > > and any "particles" in that overlapping region will probably
> > clash.
> > > > > > >
> > > > > > > Hope this helps,
> > > > > > >
> > > > > > > -Dan
> > > > > > >
> > > > > > > --
> > > > > > > -------------------------
> > > > > > > Daniel R. Roe
> > > > > > > Laboratory of Computational Biology
> > > > > > > National Institutes of Health, NHLBI
> > > > > > > 5635 Fishers Ln, Rm T900
> > > > > > > Rockville MD, 20852
> > > > > > > https://www.lobos.nih.gov/lcb
> > > > > > >
> > > > > > > _______________________________________________
> > > > > > > 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
> > > > >
> > > > _______________________________________________
> > > > 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
> >
> _______________________________________________
> 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 Mon Oct 24 2016 - 12:00:03 PDT
