RE: a bug? Re: AMBER: one more line about the npscal, box size changes and rms

From: Yong Duan <yduan.udel.edu>
Date: Thu, 7 Aug 2003 14:49:39 -0400

Answers to these questions are a bit convoluted. But, here we go ...
I personally never had a problem with either, so long as the presure
coupling factor is small.
However, for "belly" simulations when you keep part of the system
static, a few additional things need to be looked at. If the static part
of your system comprises entire molecules, npscal=1 should obviously
work better since you avoid changing bond length during presure scaling.

For the case that you only hold part of a molecule to static while allow
other parts of the same molecule to move, you probably should also use
npscal=1 to avoid changes on your static part.

In either case, please use a weak coupling constant because the
Berendsen algorithm is designed for linear regime only.

Another thing that you should keep in mind is that, because volume is
not well defined in the case when part of the system is not accounted
for, strictly speaking, presure scaling should not be allowed if the
intention is to obtain an accurate thermaldynamic ensemble. This is
because, physically, in the absence of external force (e.g., earth
gravity), presure is a consequence of spatial confinement (as ideal gas
in a box). However, in practice, this is ok for the purpose of
equilibration. But, you would have to do it carefully.

Hope this helps!

yong

-----Original Message-----
From: owner-amber.scripps.edu [mailto:owner-amber.scripps.edu] On Behalf
Of Haizhen Zhong
Sent: Thursday, August 07, 2003 1:54 PM
To: amber.scripps.edu
Subject: Re: a bug? Re: AMBER: one more line about the npscal, box size
changes and rms


Hi there,

Thanks to Yong and Andy's prompt reply.

According to my understanding from these two reply, it seems it's better
to use
npscal=1 rather than npscal=0. Since npscal = 0 is default value, there
may be some reason to use npscal=0 in certain cases and in other cases,
it
may be better to use npscal=1. Now my new question is:

"Are there some rules that define when is it best to use npscal=0 and
when
to use npscal=1? Is there any guideline in the manual?"

Or according to Yong's indication, is there a bug when npscal is set to
0?
In such case, does it mean not to use npscal = 0 at all?

Thank you so much!

Haizhen

>
> In the npscal=0, atom-centered scaling, the coordinates of each atom
are
> scaled. In the npscal=1, the COM of molecules are scaled where the
> intra-molecular geometry is unchanged.
>
> In your case, since your protein was kept static, its intra-molecular
> forces are not calculated. It actually does not move during regular
> coordinate update except at the pressure regulation stage. This should
be
> changed in the code such that the coordinates of the static part are
not
> changed during pressure regulation.
>
> yong
> On Thu, 7 Aug 2003, Haizhen Zhong wrote:
>
> > Hi there,
> >
> > I just tried the same system for water-only equilibration and all
the
> > parameters in .in file are the same except that I set npscal=1.
After
> > several ps running, the box size is shrinking. Yet the rms between
the
> > protein from the equilibration and the crystal strucutre is 0.0 and
the
> > protein looks the same as the crystal.
> >
> > So why does npscal = 0 or = 1 have such significantly influence on
the
> > protein
> > strucutre? Or is it the difference in how to present the protein
between
> > npscal = 0 or 1? If it is the presentation problem and not the
problem of
> > protein itself, is it any way to transform protein structure from
npscal =
> > 0 to the right structure as those from npscal =1, in order to have
the
> > right comparison to the crystal structure?
> >
> >
> > Thank you so much!
> >
> > Haizhen
> >
> > On Thu, 7 Aug 2003, Haizhen Zhong wrote:
> >
> > > Dear Tom and other Amber fellows,
> > >
> > > I have one question about npscal. I am using AMBER6.0. According
to the
> > > munual, npscal is default set to 0 (atom scaling). Yet when you
use this
> > > default, after the NVT ramping (water only, constant volume
heating up
> > > from 10 K to
> > > 300 K in 30 ps), during the NPT equilibration phase, as the box
size
> > > changes from (87, 87, 87) to (74, 74, 74) in an octahedral box,
the rms
> > > between the protein in the equilibration phase to the crystal also
becomes
> > > bigger and bigger. WHen the system is well equilibrated and the
box size
> > > does not change, the rms of protein to the crystal also does not
change.
> > >
> > > Now my question is: since the first ramping and equilibration is
for water
> > > only and only water molecules are in belly for allowable move,
therefore
> > > the rms between the protein and the crystal should be 0.0 for any
protein
> > > (no matter in the temperature ramping phase, or in the
equilibration
> > > phase, since it is only water allowed to move). Yet the npscal = 0
gives
> > > rms = 2.3, is it anything wrong with the protein? Or is it only
the
> > > scaling problem during the equilibration due to the change of box
size?
> > > When I compare the equilibrated protein to crystal strucutre, I
found even
> > > changes in secondary structure of protein occur? Is it right? If
it is
> > > because of npscal and scaling problem, what kind of work I have to
do to
> > > make the protein in equilibration phase comparable to the crystal
(i.e.,
> > > rms=0 for water only)?
> > >
> > > By the way, I use Sander_classic in AMBER6, and the protein is
about
> > > 200 residues. Thank you so much for your
> > > help.
> > > The .in file is as follows,
> > > &cntrl
> > > imin=0,
> > > ntx=7,
> > > scee=1.2,
> > > ntb=2,
> > > ntc=2,
> > > ntpr=50,
> > > dielc=1,
> > > irest=1,
> > > init=4,
> > > tempi=300.0,
> > > ntt=1,
> > > temp0=300.0,
> > > tautp=0.2,
> > > ntp=1,
> > > ntf=2,
> > > nstlim=50000,
> > > ntwe=50,
> > > ntwx=50,
> > > dt=.002,
> > > ndfmin=6,
> > > jfastw=0,
> > > nsnb=30,
> > > cut=12,
> > > npscal=0,
> > > ntwr=50,
> > > ntwv=0,
> > > ntwxm=0,
> > > ntwem=0,
> > > ibelly=1
> > > &end
> > > Group for belly solvent
> > > RES 206 8202
> > > END
> > > END
> > >
> > > Thank you so much,
> > >
> > > Haizhen
> > >
> > >
> > >
-----------------------------------------------------------------------
> > > The AMBER Mail Reflector
> > > To post, send mail to amber.scripps.edu
> > > To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
> > >
> > >
> >
> >
> >
-----------------------------------------------------------------------
> > The AMBER Mail Reflector
> > To post, send mail to amber.scripps.edu
> > To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
> >
>
>
>
-----------------------------------------------------------------------
> The AMBER Mail Reflector
> To post, send mail to amber.scripps.edu
> To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
>
>


-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber.scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu


-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber.scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
Received on Thu Aug 07 2003 - 19:53:01 PDT
Custom Search