AMBER Archive (2007)

Subject: Re: AMBER: TMD and "Current RMSD from reference"

From: Carlos Simmerling (carlos.simmerling_at_gmail.com)
Date: Sun Sep 23 2007 - 06:32:47 CDT


Ilyas,
no, you can't use both restraintmask and rmsmask. You
have to choose one. We felt that with restraints, there was no point in
doing a best-fit, but we thought people would use strong enough
restraints to keep the system from moving much. Have you checked
in your simulations to see how far the restrained part moves?
You can do that by runnning ptraj and calculating rmsd without
fitting. It is indeed "implicitly fitted", but if the restraints are very
weak then it may not be done well.
I suppose it could be possible in the code to do both restraints
and a best fit, but it would introduce extra complexity since we would
need to restore the coordinates after the best fit and rmsd, otherwise the
restraint energy would suffer discontinuities.

carlos

On 9/22/07, Ilyas Yildirim <yildirim_at_pas.rochester.edu> wrote:
>
> Dear Carlos,
>
> Thanks again for your response. So, you say that I can use both
> restraintmask and tgtfitmask in a targetted MD simulation, is that right?
> I did not use tgtfitmask because in the manual it says that when ntr=1 is
> set, the structures is implicitly fitted according to atoms specified in
> restraintmask. Looks like that there is a mistake in this part of the
> manual.
>
> I just searched the amber mailing list, and saw that someone already
> pointed out this. I wish I read it before running the simulations. Thanks
> again.
>
> Best,
>
> On Sat, 22 Sep 2007, Carlos Simmerling wrote:
>
> > no, that's not correct. when you use ntr=1, there is _no_ fitting
> > involved. it applies positional restraints just as normal ntr=1
> > simulations. the rmsmask specifies an rmsd calculation, but no fitting
> > is involved.
> >
> > what you are doing is certainly reasonable, but you can't expect
> > it to match something where a best fit is done.
> > if you want tgtmd simulations with best-fitting, use the fitmask.
> > this doesn't work with ntr=1.
> >
> > carlos
> >
> > On 9/22/07, Ilyas Yildirim <yildirim_at_pas.rochester.edu> wrote:
> > >
> > > Dear Carlos,
> > >
> > > Thanks for your reply, but I am not sure if what you say is right
> > > or not compared to what is written in the TMD part of the manual.
> > >
> > > What I have done in this simulation is, I put positional restraints to
> > > residues :1-2,8-11,17-18. According to the manual, the structure will
> be
> > > implicitly 'fitted' to this positional restraints (I assume that this
> > > 'fitting' is best-fitting). The targetted rmsd will be, then,
> calculated
> > > for the residues :3-7,12-16 (without doing any fitting). That is what
> I
> > > wanted to do; first best-fit the structure to residues
> :1-2,8-11,17-18,
> > > then calculate the targetted rmsd for residues :3-7,12-16.
> > >
> > > The restraint weight is good enough for the system; I put this
> positinal
> > > restraints to the terminal base pairs so that the structure will not
> fall
> > > apart during the simulation.
> > >
> > > The way I see this simulation is as follows: With these restraints,
> the
> > > structure changes according to Newtonian Mechanics. At each step,
> energy
> > > information is calculated. Also, at each step, the structure is
> > > best-fitted to the reference structure according to restraintmask.
> Then,
> > > the targetted rmsd is calculated between the reference structure and
> > > the current MD structure for the atoms defined in tgtrmsmask (without
> > > doing any fitting). This information is written in the sander output
> file
> > > as "Current RMSD from reference". I dont think that the restraint
> weight
> > > is the reason why the sander output's targetted rmsds values are
> different
> > > than the ptraj's rmsd calculation.
> > >
> > > The ptraj script that I have is as follows:
> > > ----------------- ptraj.in ---------------------
> > > reference xray.inpcrd
> > > trajin tgtmd_1.traj
> > > rms reference :1-2,8-11,17-18
> > > rms reference out rmsd_out :3-7,12-16 nofit
> > > go
> > > ------------------------------------------------
> > >
> > > The problem arises in the analysis part. When I compare the ptraj
> output
> > > with the sander output ('Current RMSD from reference' rmsd values),
> they
> > > are different. Why are they different? I assumed that the alignment
> > > protocol for targetted MD is similar to ptraj's alignment protocol.
> > > If the methods used for the alignment is different, which result am I
> > > going to use; the ptraj's results or the sander output file results?
> > >
> > > If there is something wrong with the way I look at targetted MD, I
> > > will greatly appreciate your comments. Thanks.
> > >
> > > Best,
> > >
> > > On Sat, 22 Sep 2007, Carlos Simmerling wrote:
> > >
> > > > your restraint weight is very weak.
> > > > it will calculate the non-fit rmsd for atoms in
> > > > the rmsmask, but will do no best fit.
> > > > when you use ptraj and do an alignment, then
> > > > you already have something different from what
> > > > you asked sander to do. try to use ptraj to calculate
> > > > the rmsd without any fitting at all.
> > > > alternatively, use a much stronger restraint
> > > > weight and see if the agreement in rmsd improves.
> > > > it could be something else, but I am guessing it
> > > > is the extremely weak restraint in sander as compared
> > > > to actual alignment in ptraj.
> > > >
> > > > On 9/22/07, Ilyas Yildirim <yildirim_at_pas.rochester.edu> wrote:
> > > > >
> > > > > Dear Amber Users,
> > > > >
> > > > > I am having trouble understanding how the RMSD is calculated in
> > > targetted
> > > > > MD simulations. I have the following input file:
> > > > > ------------ tgtmd.in ------------------------
> > > > > TMD: NMR->XRAY; target rmsd 3.267; resid 1-2,8-11,17-18 restrained
> > > > > &cntrl
> > > > > imin = 0, ntx = 5, nstlim = 1200000, irest=1,
> > > > > dt = 0.002, ntc = 2, ntf = 2, tol = 0.000001,
> > > > > tempi = 300.0, temp0 = 300.0,
> > > > > scee = 1.2, cut = 99.9,
> > > > > ntpr = 50, ntwx = 50, ntwr = 50,
> > > > > ntb = 0, ntt = 1, tautp = 0.5, ntp = 0,
> > > > > igb = 1, nscm = 0, nmropt = 0,
> > > > > ntr = 1, restraint_wt=0.02,
> > > > > restraintmask=":1-2,8-11,17-18",
> > > > > itgtmd = 1, tgtrmsd = 3.267, tgtmdfrc = 1.0,
> > > > > tgtrmsmask="(:3-7,12-16)",
> > > > > /
> > > > > -----------------------------------
> > > > > What this script does is it restraints the residues 1-2,8-11,17-18
> > > > > with a restraint weight of 0.02. While doing that, implicitly, it
> > > aligns
> > > > > the system with respect to restraintmask.
> > > > >
> > > > > Then, the rmsd is calculated for the residues 3-7,12-16. At each
> > > > > step, "Current RMSD from reference" is printed out. I could not
> figure
> > > out
> > > > > how this 'Current RMSD from reference' is calculated. When I
> extract
> > > that
> > > > > particular snapshot from the trajectory and use ptraj (first
> aligning
> > > the
> > > > > residues 1-2,8-11,17-18 and then calculating the rmsd of residues
> > > > > 3-7,12-16), I get pretty different result.
> > > > >
> > > > > 1. Is there a mistake the way I describe/understand the above
> input
> > > file?
> > > > > 2. Does TMD follow a different alignment algorithm?
> > > > >
> > > > > Thanks in advance for your responses.
> > > > >
> > > > > Best,
> > > > >
> > > > > --
> > > > > Ilyas Yildirim
> > > > > ---------------------------------------------------------------
> > > > > = Department of Chemistry - =
> > > > > = University of Rochester - =
> > > > > = Rochester, NY 14627-0216 - Ph.:(585) 275 67 66 (Office) =
> > > > > = http://www.pas.rochester.edu/~yildirim/ =
> > > > > ---------------------------------------------------------------
> > > > >
> > > > >
> > >
> -----------------------------------------------------------------------
> > > > > The AMBER Mail Reflector
> > > > > To post, send mail to amber_at_scripps.edu
> > > > > To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu
> > > > >
> > > >
> > >
> > > --
> > > Ilyas Yildirim
> > > ---------------------------------------------------------------
> > > = Department of Chemistry - =
> > > = University of Rochester - =
> > > = Rochester, NY 14627-0216 - Ph.:(585) 275 67 66 (Office) =
> > > = http://www.pas.rochester.edu/~yildirim/ =
> > > ---------------------------------------------------------------
> > >
> > >
> > >
> -----------------------------------------------------------------------
> > > The AMBER Mail Reflector
> > > To post, send mail to amber_at_scripps.edu
> > > To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu
> > >
> >
> >
> >
> >
>
> --
> Ilyas Yildirim
> ---------------------------------------------------------------
> = Department of Chemistry - =
> = University of Rochester - =
> = Rochester, NY 14627-0216 - Ph.:(585) 275 67 66 (Office) =
> = http://www.pas.rochester.edu/~yildirim/ =
> ---------------------------------------------------------------
>
> -----------------------------------------------------------------------
> 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