Re: Fwd: TIP4P and CHARMM27

From: Francesco Pietra (chiendarret_at_gmail.com)
Date: Thu Dec 03 2015 - 12:19:58 CST

Hi Peter:

Some of this is confusing because timestep has no effect or meaning during
> minimization.

actually, as I wrote, I used the config file about which you said "This
config file doesn’t actually minimize the system, though. “minimization on”
says that conjugate gradient minimization should be used, but unless you
actually say minimize 1000". In a TIP3P box, without restraints, with
namd2.11b1CUDA, by gradually increasing the ts from 0.01 to 0.1 to 1.0fs,
the active site conserved the experimental geometry, including my two model
cations. The system could then be gradually heated NVT to 300K and NPT
equilibrated (better say, constant rmsd for the protein) with namd2.10MIC
(at the computing center only the stable code is available) at 1atm during
3,500,000 steps at ts=1.0fs, whereby the experimental geometry of the
active site was maintained. Acceptable oscillations of bond distances by
playing the dcd file.

The same situation could be reached by first minimizing ("minimize 1000")
the system under distance colvars between my two cations. Heating and
equilibration without colvars followed as above.

I do not imply that the first method should be adopted. I run into it
accidentally, as you know, no aim to adopt it.

Why minimization without colvars expelled one of my two identical cations
(in a second trial from scratch, the same was expelled) I can not say. It
is a special model cation, may be because of that. As the work goes on, may
be I'll understand why. What makes me comfortable is that my model cation
matches the experiments, alone in a TIP3P box.

cheers
francesco

On Wed, Dec 2, 2015 at 7:51 PM, Peter Freddolino <petefred_at_umich.edu> wrote:

> Hi Francesco,
> Some of this is confusing because timestep has no effect or meaning during
> minimization.
> In general, the reliance of the simulation on such miniscule timesteps
> suggests to me something very odd with the system. Most likely, to begin
> with, it needs to be sufficiently minimized before anything else is done.
> If minimization expels the ion from the protein, then the initial ion
> placement must be pathologically bad.
> Best,
> Peter
>
> > On Dec 2, 2015, at 12:47 PM, Francesco Pietra <chiendarret_at_gmail.com>
> wrote:
> >
> > Continuing with namd2.11b1, which enforces rigid bonds on minimization,
> I noticed a drastic behavior, as follows.
> >
> > Solvating with standard TIP3P, with two of "myion" ( a special model of
> ion) at the active site of my protein, "minimize", at ts=0.1fs, expelled
> one of the ions out of the protein. Imposing colvars, the distance between
> the two increased by 30% from 3.2 to 4.0.
> >
> > In contrast, by using "run" in place of "minimize", i.e., carrying out
> MD directly, without colvars (while starting with ts=0.01fs and increasing
> it stepwise, the distance between the two ions was maintained.
> >
> > I did not try "minimize" with namd2.10 in this case.
> >
> > francesco pietra
> >
> > On Tue, Dec 1, 2015 at 10:22 AM, Francesco Pietra <chiendarret_at_gmail.com>
> wrote:
> > Hi:
> > Tried with NAMD_2.11b1_Linux-x86_64-multicore. Without any restraint,
> minimizations went on smoothly in three runs at ts = 0.1, 1.0, 1.5fs, 1000
> steps.
> >
> > Gradual heating (NVT) to 300K also went on smoothly in 20,000 steps,
> ts=1.5.fs.
> >
> > NPT had to be carried out at ts=1.0fs (at ts=1.5fs, it crashed at step
> ca 45,000 out of planned 100,000 with
> > FATAL ERROR: High global exclusion count! (1474 vs 1473) System
> unstable or pairlistdist or cutoff too close to periodic cell size.
> > RMSD reached constant values at ca half the simulation.
> >
> > Is anything that you would like to see fro the log file, or from
> elsewhere?
> >
> > francesco pietra
> >
> > On Mon, Nov 30, 2015 at 11:21 PM, Jim Phillips <jim_at_ks.uiuc.edu> wrote:
> > Hi,
> >
> > Could you try NAMD 2.11b1? The minimizer now enforces rigid bonds.
> >
> > Jim
> >
> >
> >
> > On Fri, 27 Nov 2015, Francesco Pietra wrote:
> >
> > Hello: Updating previous mail to VMD only. On restarting that work from
> > scratch, with my molecule solvated by TIP4P (Freddolino box), I was
> unable
> > to minimize the system, by any protocol of NAMD 2.10, beyond ts=0.055fs.
> > Both these and previous minimization at ts=1.2fs (this could not be
> > reproduced) did not allow heating NVT: crash at the first step from 0 to
> > 1K, because of atoms moving too fast. All those procedure are familiar
> and
> > successful in my quarters with TIP3P. Then, I came across warnings that
> > unfortunately I had missed before:
> >
> > Things get DICEY very fast if you try using ANY TIP4 models with charmm
> > forcefields, because of how deeply the water model is baked into the
> charmm
> > parameterization procedure. People do it, sometimes it seems to work
> well,
> > but there's no reason to expect that it actually SHOULD give reasonable
> > results. (Freddolino 29-Sep-2015, at 7:25 PM)
> >
> >
> > All my efforts above were prompted by having got a TIP4P-solvated atom
> > model that had been parameterized with OPLSAA in GROMACS (while I know
> how
> > to change those type of parameters to charmm27). I wonder now whether
> that
> > model was reliable work.
> >
> > cheers
> >
> > francesco pietra
> >
> > ---------- Forwarded message ----------
> > From: Francesco Pietra <chiendarret_at_gmail.com>
> > Date: Thu, Nov 26, 2015 at 10:50 PM
> > Subject: TIP4P and CHARMM27
> > To: VMD Mailing List <vmd-l_at_ks.uiuc.edu>
> >
> >
> > Hello:
> > For practical reasons I would like to use TIP4P water (Freddolino kit)
> with
> > CHARMM27, not 36 ff (I have params for ligands ready in the 27 ff).
> >
> > Water is also a ligand at the active site (starting PDB file), and that
> > also should be TIP4P. Is that conceivable? Also, should TIP4P in the
> > solvated system be constrained as indicated in the example provided by
> Prof
> > Freddolino during minimization, NVT heating, NPT? In the final productive
> > NPT I would like not to constrain any bond in the protein.
> >
> > In preliminary trials without constraining anything, minimization run
> > plainly up to ts=1.0fs; atoms moving too fast beyond that.
> >
> > Thanks for advice
> >
> > francesco pietra
> >
> >
> >
>
>

This archive was generated by hypermail 2.1.6 : Tue Dec 27 2016 - 23:21:36 CST