Re: [AMBER] Negative dihedral angle n glycan simulation

From: Irfan Alibay <irfan.alibay.lists.gmail.com>
Date: Thu, 11 Apr 2019 12:43:51 +0100

Hi,

Just to add to this, we looked into this (or at least a very similar) issue
with negative dihedral energies when running aMD simulations some time ago.
Unfortunately, I forgot to report it as it was towards the end of my PhD.
As far as I know, the aMD code will work as expected with negative dihedral
energies when using the CPU version of pmemd (do you find that this in
indeed the case for you too Rajarshi?).

However they do not in pmemd.cuda. This seems to be due to a missing
bitwise xor (checking against INT_MAX) when doing uint to double
conversions after downloading the energies from the device in the aMD
routines of gpu.cpp.
Fixing this appears to resolve the problem. I am unsure about GaMD, but it
appears that a similar thing might also be happening.

I can probably create a diff patch for it, if more code details are needed.

Many thanks,

Irfan Alibay
Postdoctoral Research Associate
Department of Biochemistry
The University of Oxford



On Wed, Apr 10, 2019 at 8:14 PM Lachele Foley <lf.list.gmail.com> wrote:

> Replying to the list so this info is more generally available in case
> others wonder.
>
> The files look ok best I can tell without digging really deeply. The
> GLYCAM force field does have some negative amplitudes in the dihedral
> terms. So, it is certainly possible to have negative dihedral values.
> This is a small effect, and the presence of protein or other residues
> usually make it so that the overall values are negative. But, this
> file is just an oligosaccharide in water. Water has no dihedrals, so
> no dihedral terms. And, the specific oligosaccharide contains some of
> the atom types that can have negative values in their associated
> dihedrals. I didn't check every dihedral individually (because that's
> a lot of work), but I think it's ok that you have negative DIHED
> values. If you're really worried, try using parmtop utilities in
> AmberTools too look closely at all the dihedrals. Those tools work
> well.
>
> Maybe ask the gamd folks if they can make it possible for negative
> dihedral energies not to cause trouble?
>
> It might not be a problem anyway. Consider the amount of the effect
> compared to what gamd does. If you are trying to increase sampling,
> you're probably already messing with the energy function. If the
> DIHED values are small compared to other bonded energies, it might not
> make a lot of difference to the outcome, depending on exactly what
> gamd is doing. Keep in mind that dihedral energies are just
> corrections to the other bonded terms, at least in the GLYCAM
> philosophy.
>
> On Tue, Apr 9, 2019 at 1:32 PM Lachele Foley <lf.list.gmail.com> wrote:
> >
> > Send it to my personal email (lf.list.gmail.com).
> >
> > On Tue, Apr 9, 2019 at 3:56 AM Rajarshi Roy <phd1701171011.iiti.ac.in>
> wrote:
> > >
> > > Dear Mam,
> > > Did you receive the topology file? In amber mailing list is not
> allowing me
> > > to send those files because of the size issue.
> > >
> > > On Tue, Apr 9, 2019 at 9:13 AM Rajarshi Roy <phd1701171011.iiti.ac.in>
> > > wrote:
> > >
> > > > Dear Mam,
> > > > I am not still able to figure it out. I am simulating a proteoglycan
> chain
> > > > only. So there is no protein part. My chain just consists of dermatan
> > > > sulphate only. Please check my topology and coordinate files. I am
> not able
> > > > to find any error or discrepancy in output files or by visualization.
> > > > Thank you in advance
> > > >
> > > > On Mon, Apr 8, 2019 at 9:20 PM Lachele Foley <lf.list.gmail.com>
> wrote:
> > > >
> > > >> Also... I'm pretty sure they shouldn't be negative. So, I might
> need
> > > >> a lot of details.
> > > >>
> > > >> If you already figured it out, just let us know.
> > > >>
> > > >> On Mon, Apr 8, 2019 at 11:24 AM Lachele Foley <lf.list.gmail.com>
> wrote:
> > > >> >
> > > >> > Sorry to take so long to respond. There's lots going on.
> > > >> >
> > > >> > Can you tell me more molecular details about your system?
> > > >> > Specifically, how large is the 'proteo' part and how large is the
> > > >> > 'glycan' part? Feel free to send the input prmtop and inpcrd to
> me
> > > >> > (to just me is fine). Thanks!
> > > >> >
> > > >> > On Tue, Apr 2, 2019 at 7:05 AM Rajarshi Roy <
> phd1701171011.iiti.ac.in>
> > > >> wrote:
> > > >> > >
> > > >> > > Dear AMBER user and developers,
> > > >> > > I am trying to simulate a proteoglycan structure using
> conventional
> > > >> > > molecular dynamics technique. Everything seems to be fine in the
> > > >> output
> > > >> > > file where only the dihedral energy is showing a negative
> value. So
> > > >> why am
> > > >> > > I getting the negative value in dihedral energy? Is there any
> > > >> discrepancy
> > > >> > > in my system?
> > > >> > > Secondly, when I am trying to simulate the same system using
> amd or
> > > >> gamd in
> > > >> > > dual boost mode, but dihedral boost potential remains zero
> throughout
> > > >> our
> > > >> > > simulation. As discussed in the gamd archive it is because of
> the
> > > >> negative
> > > >> > > dihedral energy.
> > > >> > > So is it an error to get negative dihedral energy or can I use
> the
> > > >> single
> > > >> > > potential boost in amd or gamd technique?
> > > >> > > It will be beneficial for me if you clarify all these doubts.
> > > >> > > Thank you in advance!
> > > >> > >
> > > >> > > --
> > > >> > > *with regards*
> > > >> > >
> > > >> > > *Rajarshi Roy*
> > > >> > >
> > > >> > >
> > > >> > > *PhD Research Scholar*
> > > >> > >
> > > >> > > *Biosciences and Biomedical Engineering*
> > > >> > >
> > > >> > > *Indian Institute of Technology, Indore*
> > > >> > > *India*
> > > >> > > _______________________________________________
> > > >> > > AMBER mailing list
> > > >> > > AMBER.ambermd.org
> > > >> > > http://lists.ambermd.org/mailman/listinfo/amber
> > > >> >
> > > >> >
> > > >> >
> > > >> > --
> > > >> > :-) Lachele
> > > >> > Lachele Foley
> > > >> > CCRC/UGA
> > > >> > Athens, GA USA
> > > >>
> > > >>
> > > >>
> > > >> --
> > > >> :-) Lachele
> > > >> Lachele Foley
> > > >> CCRC/UGA
> > > >> Athens, GA USA
> > > >>
> > > >> _______________________________________________
> > > >> AMBER mailing list
> > > >> AMBER.ambermd.org
> > > >> http://lists.ambermd.org/mailman/listinfo/amber
> > > >>
> > > >
> > > >
> > > > --
> > > > *with regards*
> > > >
> > > > *Rajarshi Roy*
> > > >
> > > >
> > > > *PhD Research Scholar*
> > > >
> > > > *Biosciences and Biomedical Engineering*
> > > >
> > > > *Indian Institute of Technology, Indore*
> > > > *India*
> > > >
> > >
> > >
> > > --
> > > *with regards*
> > >
> > > *Rajarshi Roy*
> > >
> > >
> > > *PhD Research Scholar*
> > >
> > > *Biosciences and Biomedical Engineering*
> > >
> > > *Indian Institute of Technology, Indore*
> > > *India*
> > > _______________________________________________
> > > AMBER mailing list
> > > AMBER.ambermd.org
> > > http://lists.ambermd.org/mailman/listinfo/amber
> >
> >
> >
> > --
> > :-) Lachele
> > Lachele Foley
> > CCRC/UGA
> > Athens, GA USA
>
>
>
> --
> :-) Lachele
> Lachele Foley
> CCRC/UGA
> Athens, GA USA
>
> _______________________________________________
> 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 Thu Apr 11 2019 - 05:00:02 PDT
Custom Search