AMBER Archive (2007)

Subject: Re: AMBER: hybrid remd

From: Carlos Simmerling (carlos.simmerling_at_gmail.com)
Date: Thu May 10 2007 - 08:39:43 CDT


you're right- in multisander.f there is a parameter "size" that is 10,000.
it should be checked against system size but isn't, we'll have to change
it to be dynamically allocated. For now you can just increase it to 20k and
recompile.

I should point out that your systems looks way bigger than anything we've
tested,
so you'll need to be careful. REMD on something that big is really a
research
problem, you are likely to still have huge sampling problems. I guess what
I mean is that I hope you are really familiar with REMD and standard
simulations
before doing this.
carlos

On 5/10/07, Geoff Wood <geoffrey.wood_at_epfl.ch> wrote:
>
> Hi Carlos,
>
> Yes I have, they are different and irregular. By that I mean the restart
> files for different replicas have different numbers of atoms specified.
>
> I have been able to get the my hybrid calculation going by reducing the
> number of kept waters from 350 to 50 and also the total number of waters
> from approx 14,000 to approx 9000. So i was wondering if there was an
> array-size issue somewhere?
>
>
> Dr Geoffrey Wood
> Ecole Polytechnique Fédérale de Lausanne
> SB - ISIC - LCBC
> BCH 4108
> CH - 1015 Lausanne
>
>
>
>
> On May 10, 2007, at 1:20 PM, Carlos Simmerling wrote:
>
> have you checked the number of atoms in your strip prmtop and
> compared to the # of atoms sander is producing in the .strip restart
> files?
>
> On 5/9/07, Geoff Wood <geoffrey.wood_at_epfl.ch> wrote:
> >
> > Hi Carlos,
> > Thanks for clearing up that issue for me. I just wanted to return to
> > the original problem I was having. For my system, I have a full periodic
> > set of files and a set of files for the stripped non-periodic system. I
> > also have a coordinate file for the stripped system, since I created the
> > parm and the corresponding crd files in leap.
> >
> > If I forget about running the hybrid calculation and run a standard remd
> > simulation on the full system then it runs without problems.
> > Alternatively, if I run a standard remd run on the reduced system then it
> > also runs without problems. This suggests that there is no problems with my
> > crd or parm files in both cases.
> >
> > However, when I attempt to run the hybrid calculation then I get the
> > error that I reported. From our correspondence would you be able to shed
> > light on why this may be happening?
> >
> > Cheers,
> >
> > Dr Geoffrey Wood
> > Ecole Polytechnique Fédérale de Lausanne
> > SB - ISIC - LCBC
> > BCH 4108
> > CH - 1015 Lausanne
> >
> >
> >
> >
> > On May 9, 2007, at 11:18 PM, Carlos Simmerling wrote:
> >
> > exactly. you need to be careful because they may not actually have full
> > solvent shells.
> > also you need to make sure you keep enough water to solvate both
> > solutes.
> >
> > On 5/9/07, Geoff Wood <geoffrey.wood_at_epfl.ch> wrote:
> > >
> > > Hi Carlos,
> > > Thankyou for your response to my questions. I am still slightly
> > > confused (its just not only sander) so I just wanted to clarify what you
> > > have said. Sander should be able to run the simulation with multiple
> > > solutes but it is unclear how meaningful the results will be?
> > >
> > > Dr Geoffrey Wood
> > > Ecole Polytechnique Fédérale de Lausanne
> > > SB - ISIC - LCBC
> > > BCH 4108
> > > CH - 1015 Lausanne
> > >
> > >
> > >
> > >
> > > On May 9, 2007, at 10:52 PM, Carlos Simmerling wrote:
> > >
> > > Hi Geoff, yes that's true, I think that it's not so much that sander
> > > will get confused as it might
> > > cause problems in the centering/imaging and possible inability to get
> > > a single set of
> > > coordinates with a complete solvent shell around multiple solutes.
> > > This isn't a problem in the
> > > fully solvated system during MD due to the periodicity, but the hybrid
> > > calculation
> > > is done without periodicity. As far as ions go, I think one needs to
> > > have the same
> > > caution as with explicit counterions and any implicit solvent- we're
> > > not sure how well
> > > they work or what the limitations are, so treat it cautiously and only
> > > try it after you're
> > > sure you understand what you are doing without them.
> > > carlos
> > >
> > > On 5/9/07, Geoff Wood < geoffrey.wood_at_epfl.ch> wrote:
> > > >
> > > > Hello Asim,
> > > >
> > > > Thank you very much for your suggestions they were very helpful. I
> > > > have made sure that the topology and number of waters correspond to what
> > > > sander is expecting. That being said, I haven't been able to make much
> > > > progress. If I understand your email correctly then any attempt to run
> > > > hybrid remd with counter-ions will cause sander to get confused as to what
> > > > is solute and solvent? If this is the case, then would the same be true if
> > > > I decided to run a hybrid remd run with more than one solute (say a dimer of
> > > > two proteins)?
> > > >
> > > > Thanks in advance,
> > > >
> > > > Dr Geoffrey Wood
> > > > Ecole Polytechnique Fédérale de Lausanne
> > > > SB - ISIC - LCBC
> > > > BCH 4108
> > > > CH - 1015 Lausanne
> > > >
> > > >
> > > >
> > > >
> > > > On May 9, 2007, at 9:07 PM, Asim Okur wrote:
> > > >
> > > > Hi,
> > > >
> > > > Before running Hybrid Solvent REMD you should read the manual and
> > > > the
> > > > article (Okur, Roe, Cui, Hornak and Simmerling, JCTC, 2007)
> > > > especially
> > > > the warnings in the manual (Page 168).
> > > >
> > > > Your hybrid topology file should have exactly the same number of
> > > > waters as the reduced system generated by sander. Use of
> > > > counterions
> > > > has not been implemented/tested with hybrid remd and should not be
> > > > used.
> > > >
> > > > Also, I see in your GB calculations (mdin.rep01.strip) you set
> > > > nstlim
> > > > = 1000. You should not do any dynamics during the GB step. Just
> > > > single
> > > > step energy evaluation is needed. You can see examples of input
> > > > files
> > > > in the remd and hybrid remd test cases in amber9.
> > > >
> > > > I hope this helps,
> > > >
> > > > Asim
> > > >
> > > > --
> > > > Asim Okur, Ph.D.
> > > > Stony Brook University
> > > > Chemistry Department
> > > > Stony Brook, NY 11733
> > > > (631) 632 1560
> > > >
> > > >
> > > > On 5/9/07, Geoff Wood < geoffrey.wood_at_epfl.ch> wrote:
> > > >
> > > >
> > > >
> > > >
> > > > Slightly from first email but essentially the same:
> > > >
> > > >
> > > > To whom it may concern,
> > > >
> > > > I am attempting to run hybrid remd simulations with amber 9 but I
> > > > find that
> > > > I cannot get them started. I am probably making a simple mistake
> > > > in the
> > > > specification of the file names, any help would be much appreciated.
> > > >
> > > > The error I am getting is the following:
> > > >
> > > >
> > > > --------------------------------------------------------------------------------
> > > > 3. ATOMIC COORDINATES AND VELOCITIES
> > > >
> > > > --------------------------------------------------------------------------------
> > > >
> > > >
> > > > FATAL: NATOM mismatch in coord and topology files
> > > >
> > > >
> > > > This error appears in the stripped output file (i.e. a file called
> > > > something
> > > > like mdout.rep01.strip)
> > > > In the full output file it looks as if the calculation has finished
> > > > (the
> > > > file is called mdout.rep01) i.e.
> > > >
> > > > Exiting runmd after getting initial energies for REMD 0
> > > > wrapping first mol.: 1.009557807795775E-014 68.4142020000000
> > > > 96.4591000000000
> > > >
> > > >
> > > > --------------------------------------------------------------------------------
> > > > 5. TIMINGS
> > > >
> > > > --------------------------------------------------------------------------------
> > > >
> > > > |>>>>>>>>PROFILE of Average TIMES>>>>>>>>>
> > > >
> > > >
> > > >
> > > >
> > > >
> > > > The groupfile looks like
> > > >
> > > > -O -i mdin.rep01 -o mdout.rep01 -p prm -c inpcrd.rep01 -r
> > > > restrt.rep01 -x
> > > > mdcrd.rep01 -inf mdinfo.rep01
> > > > .
> > > > .
> > > > .
> > > > .
> > > >
> > > > The input files look like:
> > > >
> > > > Full box (mdin.rep01)
> > > >
> > > > comment
> > > > &cntrl
> > > > imin = 0, irest = 1, ntx = 2,
> > > > ntb = 2, pres0 = 1.0, ntp = 1,
> > > > taup = 2.0,
> > > > cut = 10, ntr = 0,
> > > > ntc = 2, ntf = 2,
> > > > tempi = 0.0, temp0 = 300.0,
> > > > ntt = 3, gamma_ln = 1.0,
> > > > repcrd=0, numexchg=40, numwatkeep=350,
> > > > nstlim = 1000, dt = 0.002,
> > > > ntpr = 100, ntwx = 100, ntwr = 1000
> > > > &end
> > > > &ewald
> > > > &end
> > > >
> > > > Reduced number of waters ( mdin.rep01.strip)
> > > >
> > > > comment
> > > > &cntrl
> > > > imin = 0, irest = 1, ntx = 2,
> > > > ntb = 0,
> > > > igb=1,
> > > > cut = 999, ntr = 0,
> > > > ntc = 2, ntf = 2,
> > > > tempi = 0.0, temp0 = 366.0,
> > > > ntt = 3, gamma_ln = 1.0,
> > > > repcrd=0, numexchg=40, numwatkeep=350,
> > > > nstlim = 1000, dt = 0.002,
> > > > ntpr = 100, ntwx = 100, ntwr = 1000
> > > > /
> > > >
> > > >
> > > > I have placed both the full parm file (prm) and reduced parm file
> > > > (prm.strip) in the same directory. For the stripped parm I have
> > > > tried with
> > > > and without the counter ions that are present in the full
> > > > calculation. I
> > > > think it is preferable to omit them in a GB energy evaluation. The
> > > > stripped
> > > > parm was made using leap and a pdb that ptraj generated from a
> > > > short md run
> > > > involving the full system.
> > > >
> > > > The input coordinates are for the full system ( from a short md run
> > > > for
> > > > equilibration) and are placed in the same directory as inpcrd.rep01etc.
> > > >
> > > > Thanks in advance,
> > > >
> > > >
> > > > Dr Geoffrey Wood
> > > > Ecole Polytechnique Fédérale de Lausanne
> > > > SB - ISIC - LCBC
> > > > BCH 4108
> > > > CH - 1015 Lausanne
> > > >
> > > >
> > > >
> > > >
> > > >
> > > >
> > > >
> > > > -----------------------------------------------------------------------
> > > > The AMBER Mail Reflector
> > > > To post, send mail to amber_at_scripps.edu
> > > > To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu
> > > >
> > > >
> > > >
> > >
> > >
> >
> >
> >
>

-- 
===================================================================
Carlos L. Simmerling, Ph.D.
Associate Professor                 Phone: (631) 632-1336
Center for Structural Biology       Fax:   (631) 632-1555
CMM Bldg, Room G80
Stony Brook University              E-mail: carlos.simmerling_at_gmail.com
Stony Brook, NY 11794-5115          Web: http://comp.chem.sunysb.edu
===================================================================

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