AMBER Archive (2007)

Subject: Re: AMBER: Fwd: a request

From: Gustavo Seabra (gustavo.seabra_at_gmail.com)
Date: Sat Jan 27 2007 - 22:52:08 CST

Hi,

I don't know the specifics of your system, nor the details on how you
intend to simulate it. The protocol you mention seems to be OK for me.
The important point is for you to remember that when you go to QM/MM
you will be using a different hamiltonian, so a system that was in
equilibrium in a classical hamiltonian might not be in equilibrium for
the QM/MM case. So, my suggestion is that:

1. The first part of the equilibration, when youo are just trying to
remove bad contacts, bring the system to room temperature, adjust
pressure/volume, etc., you should do in pure MM. Besides being faster,
it's more stable. You don't care much about the results anyway, since
it's not production.

2. Once the system is close to what you want for production, turn on
QM/MM. BUT, since the hamiltonian changed, you need to relax tour
system a bit before running production.

The specific details of the protocol depend on the systes, so I let
that for you to decide.

Good luck,

Gustavo.

On 1/27/07, santanu roy <66.santanu_at_gmail.com> wrote:
> ok, just correct me if I have understood you properly, after minimization
> iof the whole system is done I will take the system to a temperature say
> 300 k at constant volume condition using MM only for say 30 ps, then I will
> remove the constant volume condition and allow the system to relux and
> constant pressure condition to switch on, and this constant temperature &
> pressure condition along with the relaxation of the system will be
> carried out bye qmmm method for say 300 ps
>
> I hope this is what you meant.
> Santanu
>
>
> On 1/25/07, Gustavo Seabra <gustavo.seabra_at_gmail.com> wrote:
> >
> > You are welcome.
> >
> > But before you do that, I just wanted to clarify one point: before you
> > use qm/mm for production, you do need to relax the system for a while,
> > using qm/mm. You will find that, even if you equilibrate the system
> > using MM first, when you change to qm/mm the system will most likely
> > *not* be in equlibrium, and you'll need to relax it (with qm/mm)
> > *before* doing any qm/mm production run.
> >
> > Good luck,
> >
> > Gustavo.
> >
> > On 1/25/07, santanu roy <66.santanu_at_gmail.com> wrote:
> > > Dear Sir,
> > > Thank you very much for your reply. Before getting your
> reply I
> > > solved the problem, actually
> > > it was the large no. of atoms within the qm part , which made the
> problem,
> > > saying infinite formate loop,
> > > When I discarded some atoms, I got the result.
> > >
> > > But I will follow the way you have showed me, I will use qmmm method
> during
> > > productive run only.
> > > That could be the best thing.
> > > Thanks again
> > > Santanu
> > >
> > > On 1/25/07, Gustavo Seabra <gustavo.seabra_at_gmail.com > wrote:
> > > >
> > > > Hi,
> > > >
> > > > As for the atom numbers, your scheme seems OK. The numbers you get
> > > > from the PDB file should be correct. You can also try the 'writepdb'
> > > > optioin in the qm/mm namelist, that will give you a pdb file with just
> > > > the QM region, so you can check your system.
> > > >
> > > > Note that you can also use 'QMMASK' to set the QM region. That may be
> > > > easier than listing atom-by-atom.
> > > >
> > > > Now, there's one thing I would *not* recommend you to do, and it is
> > > > running the first minimization (just to remove bad contacts) already
> > > > with QM. At leas in my experience, you may run into lots of
> > > > convergence problems there. You should first minize the system using
> > > > just MM, then try to take it to the desired temperature first, all
> > > > using MM. AFTER that, you can change part of the system to QM, and
> > > > then relax the system for a while more. This procedure seems to work
> > > > best for me. (Ross, comments?)
> > > >
> > > > Gustavo.
> > > > P.S. If you are running into problems with putting all files in gmail,
> > > > you can try to tar all files together and send just one tar file.
> > > >
> > > > On 1/25/07, santanu roy < 66.santanu_at_gmail.com> wrote:
> > > > >
> > > > > Sir, I could not send prmtop and inpcrd file, gmail problem.
> > > > > looking forward for your reply.
> > > > >
> > > > >
> > > > > ---------- Forwarded message ----------
> > > > > From: santanu roy < 66.santanu_at_gmail.com>
> > > > > Date: Jan 25, 2007 12:42 PM
> > > > > Subject: Re: a request
> > > > > To: Ross Walker <ross_at_rosswalker.co.uk >
> > > > >
> > > > >
> > > > >
> > > > > Dear Sir,
> > > > > Let me tell you what I am actually doing. I am
> using
> > > AMBER
> > > > > qmmm method for simulating Thermolysin. Some part of the Active site
> > > > > has been kept into the QM region and other part obviously moves into
> the
> > > > > MM region. I , atfirst, using Xleap solvated the system of
> interest
> > > i.e.
> > > > > thermolysin using solvatebox command. Then I made
> thermolysin.prmtop
> > > > > and thermolysin.inpcrd file. Then using ambpdb command I made
> > > > > solvated_thermolysin.pdb , i.e. using parameter and coordinate file
> and
> > > > > ambpdb command I have got a new pdb file of solvated thermolysin.
> > > > >
> > > > > Now, I have to minimise the solvated thermolysin, so atfirst my job
> > > would
> > > > > be to minimise the water solvant keeping thermolysin fixed. There I
> > > found
> > > > > the problem. in iqmatoms=......, I took the atom no. from the
> newly
> > > > > formed solvated_thermolysin.pdb. I have a doubt over here, as when
> we
> > > > > run the command for minimisation only prmtop and inpcrd file
> information
> > > > > are given, so if we use atom no. in qmmm key , taken from
> > > > > Solvated_thermolysin.pdb which is made from prmtop and inpcrd file,
> > > > > will that be correct? Actually there is no other way to assign atom
> no.
> > > in
> > > > > qmmm key I have found. I think I am right in this case.
> > > > >
> > > > > I am attaching the input and output file, and also prmtop & inpcrd
> > > file,
> > > > > solvated_thermolysin.pdb file.
> > > > >
> > > > > I am using SGI ALTIX, MPIRUN, sometimes using 8 parallel processor
> > > > > and sometimes 4 parallel processor.
> > > > >
> > > > >
> > > > >
> > > > >
> > > > >
> > > > >
> > > > > On 1/24/07, Ross Walker < ross_at_rosswalker.co.uk > wrote:
> > > > > >
> > > > > >
> > > > > > Hi Santanu,
> > > > > >
> > > > > > The input format for QMMM with link atoms is identical to QMMM
> without
> > > > > link atoms. Sander takes care of all the link atom treatment for you
> > > > > automatically. The best is if you can send me the calculation you
> are
> > > > > running so I can take a look at it. Please send me the mdin, prmtop
> and
> > > > > inpcrd files as well as the output file where you are having
> problems.
> > > > > >
> > > > > > It would also be useful to know what system (OS, processor etc)
> you
> > > are
> > > > > running this on, what compilers you used. Is this serial or
> parallel?
> > > And
> > > > > which bugfixes have been applied to the Amber installation?
> > > > > >
> > > > > > All the best
> > > > > > Ross
> > > > > >
> > > > > > /\
> > > > > > \/
> > > > > > |\oss Walker
> > > > > >
> > > > > > | HPC Consultant and Staff Scientist |
> > > > > > | San Diego Supercomputer Center |
> > > > > > | Tel: +1 858 822 0854 | EMail:- ross_at_rosswalker.co.uk |
> > > > > > | http://www.rosswalker.co.uk | PGP Key available on request |
> > > > > >
> > > > > > Note: Electronic Mail is not secure, has no guarantee of delivery,
> may
> > > not
> > > > > be read every day, and should not be used for urgent or sensitive
> > > issues.
> > > > > >
> > > > > >
> > > > > >
> > > > > > ________________________________
> > > > > From: santanu roy [mailto:66.santanu_at_gmail.com]
> > > > > > Sent: Wednesday, January 24, 2007 01:11
> > > > > > To: Ross Walker
> > > > > > Subject: Re: a request
> > > > > >
> > > > > >
> > > > > >
> > > > > >
> > > > > >
> > > > > > Dear Sir,
> > > > > > Thank yoy very much for your reply. I have got my
> > > answer.
> > > > > > But I am facing problem while silmulating qmmm system.
> > > > > >
> > > > > > Can you please send me the input formate for qmmm with link atom
> > > method.
> > > > > Neither in the tutorial , nor in the manual the qmmm formate
> including
> > > link
> > > > > atom is given. I myself tried out, but getting some error like
> > > > > > 'aborting' .
> > > > > >
> > > > > > I will be highly obliged if you kindly respond.
> > > > > > Thanking you,
> > > > > > Santanu
> > > > > >
> > > > > >
> > > > > >
> > > > > > On 1/17/07, Ross Walker <ross_at_rosswalker.co.uk > wrote:
> > > > > > >
> > > > > > >
> > > > > > > Dear Santanu,
> > > > > > >
> > > > > > > I'm not entirely sure I understand your question correctly.
> However,
> > > I
> > > > > assume you are refering to the first amber tutorial on modelling
> DNA.
> > > Here
> > > > > we start doing gas phase simulations where we run two simulations,
> one
> > > > > without a cutoff and one with. The key point here is that the cutoff
> > > > > simulation gives a stable system but the no-cutoff simulation does
> not.
> > > Thus
> > > > > one could make the "wrong" conclusion that the cutoff sim is giving
> the
> > > > > correct result. However, these simulations are in a vacuum at 300K.
> If I
> > > > > really took a 10mer of DNA and heated it up to 300K in a vacuum it
> would
> > > not
> > > > > be stable - the two chains would alsmost certainly come apart and
> the
> > > > > molecule would decompose. Thus the no-cutoff simulation is probably
> > > correct.
> > > > > > >
> > > > > > > The important point is that using a cutoff is 'never' a good
> idea.
> > > You
> > > > > can get away with something liek 16 angstroms or more in an implicit
> > > solvent
> > > > > GB simulation as there is a dielectric screening factor that reduces
> > > things
> > > > > by a factor of 80. However, realisticlly here one should not use a
> > > cutoff if
> > > > > one can afford it computationally. Electrostatic interactions are so
> > > long
> > > > > ranged that you have to go to 30 or 40 angstroms before you get down
> to
> > > > > noise.
> > > > > > >
> > > > > > > However, with PME doing periodic boundaries the situation is
> > > slightly
> > > > > different. You should probably read up on thePME method from the
> papers
> > > > > cited in the Amber manual and make sure you understand the
> mathematics.
> > > > > Essentially what you do is take what is a divergent (infinite) sum
> of
> > > pair
> > > > > wise interactions in real space and cast it into a convergent sum in
> > > > > reciprocal space. You then use an FFT to calculate the reciprocal
> space
> > > > > contribution. You still have a cutoff of say 8 angstroms or so to
> > > account
> > > > > for VDW interactions and also for load balancing issues. This way 8
> > > > > angstroms worth of electrostatics is done in the direct space sum
> within
> > > the
> > > > > cutoff while the remainder is done in reciprocal space and adjusted
> by
> > > an
> > > > > erfc function (error function) to go to zero at the cutoff. In this
> way
> > > PME
> > > > > gives you effectively infinite electrostatics without requiring you
> to
> > > > > calculate infinite pairwise interactions. Hence why you can
> essentially
> > > use
> > > > > a cutoff with periodic boundaries without problems - as LONG AS YOU
> ARE
> > > > > USING PME.
> > > > > > >
> > > > > > > I hope this helps explain things. Note if you further questions
> it
> > > is
> > > > > probably best to post them to the Amber mail reflector - see
> > > > > http://amber.scripps.edu for details on how to subscribe.
> > > > > > >
> > > > > > > All the best
> > > > > > > Ross
> > > > > > >
> > > > > > > /\
> > > > > > > \/
> > > > > > > |\oss Walker
> > > > > > >
> > > > > > > | HPC Consultant and Staff Scientist |
> > > > > > > | San Diego Supercomputer Center |
> > > > > > > | Tel: +1 858 822 0854 | EMail:- ross_at_rosswalker.co.uk |
> > > > > > > | http://www.rosswalker.co.uk | PGP Key available on request |
> > > > > > >
> > > > > > > Note: Electronic Mail is not secure, has no guarantee of
> delivery,
> > > may
> > > > > not be read every day, and should not be used for urgent or
> sensitive
> > > > > issues.
> > > > > > >
> > > > > > >
> > > > > > >
> > > > > > > ________________________________
> > > > > From: santanu roy [mailto: 66.santanu_at_gmail.com]
> > > > > > > Sent: Friday, January 12, 2007 08:31
> > > > > > > To: amber_tutorial_query_at_rosswalker.co.uk ;
> > > > > ross_at_rosswalker.co.uk
> > > > > > > Subject: a request
> > > > > > >
> > > > > > >
> > > > > > >
> > > > > > > Dear Sir,
> > > > > > > I have gone through almost all of the amber
> > > tutorial,
> > > > > and tested it.
> > > > > > >
> > > > > > > I have some basic questions as the following
> > > > > > > 1) why we generally take cutoff as 8 to 10 angstrom? in this
> case
> > > we
> > > > > get actually correct result, why? in case of no cutoff we get
> > > > > > > worst result. why?
> > > > > > > is it like VDW interaction, electrostatic interaction get very
> less
> > > > > value, so we should ignore those interaction after that cut off?
> > > > > > > and in case of no cut off we consider those interactions, is it
> the
> > > > > reason for unstable result?
> > > > > > >
> > > > > > > 2) does cut off affect on periodic boundary condition?
> > > > > > >
> > > > > > > I am looking forward for your answer.
> > > > > > > Regards
> > > > > > > Santanu Roy
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu