> Thanks for the refernce My mdin files (in sequence)
I think Prof Duan's advice is salient, but here is
some commentary on your files that indicates a fix you could
try at the end.
> minimization of water or solvent
> molecules
> &cntrl
> imin=1,
> ntb=1,
> cut=10.0,
> maxcyc=100,
> ncyc=10,
> ntt=0,
> ntr=1,
> &end
>
> Hold the monomers
> fixed
> 1.0
>
> RES 1
> 85
> END
>
> END
Have you visualised the result of this?
100 steps of minimization is a small step to take. It is much better
than nothing, but if you do 1000 steps restrained and unrestrained in
parallel, and compare the results, I am not sure if you'd see much
difference.
>
> minimization of solvent and solute
> &cntrl
> imin=1,
> ntb=1,
> cut=10.0,
> maxcyc=100,
> ncyc=10,
> ntt=0,
> &end
It runs so fast that maxcyc=1000 would not hurt. That said, you could
be lucky and need no minimization.
>
> heat/dynamics at constant
> volume
> &cntrl
> imin=0,
> ntxo=1,
> ntrx=1,
> cut=10.0,tempi=0.1,
> ntpr=500, ntwx=500,
> ntwe=500,
> nstlim=1000,
Note that this nstlim may be the controlling one, sorry I didn't view
the output, but wondering if the warming protocol actually ran as
specified below:
> temp0=50.0,
>
> dt=0.0005,
> ntc=1, ntf=1,
Reconsider ntc=2:
A random search result out of context (2008):
"Short answer NTC=2 with NTF=1 or NTF=2 should be fine in both cases. "
http://structbio.vanderbilt.edu/archives/amber-archive/2008/1981.php
> tol=0.00001,
> ntx=1,
> irest=0,
> ntb=1,tautp=1,
> ntt=1,
> nmropt=1,ntr=1
> /
>
> &wt
> type='TEMP0', istep1=0, istep2=1000, value1=0.0,
> value2=50.0,
> /
> &wt
> type='TEMP0', istep1=1001,istep2=80000,value1=50.0,
> value2=50.0,
> /
>
> &wt
> type='END',
> /
>
> GROUP FOR
> CONSTRAINTS
> 1.0
>
> RES 1
> 85
> END
>
> END
>
> heat/dynamics at constant
> volume
> &cntrl
> imin=0,
> ntxo=1,
> ntrx=1,
> cut=10.0,tempi=50,
> ntpr=500, ntwx=500,
> ntwe=500,
> nstlim=1000,
Again nstlim is small. In this case the rest of the mdin is
consistent, it is just such a short time to run, especially at
so small a time step. I strongly recommend viewing such trajectories
to get a feel for how equilibration is progressing. If you can get
a 100,000 step trajectory generated quickly (e.g. using my favorite,
100K vacuum equilibration), it is a sight to see. Once this is done,
you can see by the movement whether there are any internal (real or
ff artifact) distortions in the initial model. Water will modulate
that result considerably, of course, and the visualisation with
water is an enterprise in its own right, but for the price vacuum
results are instructive.
> temp0=50.0,
> dt=0.0005,
> ntc=1, ntf=1,
> tol=0.00001,
> ntx=1,
> irest=0,
> ntb=1,tautp=1,
> ntt=1,
> nmropt=1,ntr=1
> /
> &wt
> type='TEMP0', istep1=0, istep2=100, value1=50.0,
> value2=50.0,
> /
> &wt
> type='TEMP0', istep1=101,istep2=1000,value1=50.0,
> value2=50.0,
> /
> &wt
> type='END',
> /
>
> GROUP FOR
> CONSTRAINTS
> 1.0
>
> RES 1
> 85
> END
>
> END
Then you minimise again? This discards velocities.
> minimization
> &cntrl
>
> imin=1, ntx=1, drms=0.0001,
> ntmin=1,
> ntrx=1, ntxo=1,
> nscm=100,
> ntb=1,
> ntf=1,
> irest=0,
> cut=10.0,
> ntpr=100, ntwx=100,
> ntwe=100,
> nsnb=10,
> tol=0.0005,
> ntc=1,
> maxcyc=10,
> ncyc=100,
> &end
>
If the next is what blew up, my guess is that it is because you are setting
temperature on the result of a minimization, which has no velocities.
> constant pressure
> equilibration
> &cntrl
>
> imin=0,
> ntx=1,
> irest=0,
> ntxo=1,tempi=50.0,
> ntpr=1000, ntwx=1000,
> ntwe=1000,
> nstlim=10000,tautp=2,
> temp0=50.0,
> dt=0.001,
> nscm=100,
>
> igb=0,ntb=2,ntp=1,
> ntt=1,
> nsnb=20,ntr=1,
> &end
>
> GROUP FOR
> CONSTRAINTS
> 1.0
>
> RES 1
> 85
> END
>
> END
>
I suggest running this with irest=1 on the result of your longer
constant volume equilibration, skipping minimization. See comments
above on where earlier runs may need re-doing.
Bill
> Thanks
> Kirtana
>
> On Sun, Dec 4, 2011 at 3:36 PM, Bill Ross <ross.cgl.ucsf.edu> wrote:
>
> > > I attached the output files in my earlier reply.
> >
> > I rarely open attachments. Mdin files are small and worth comparing
> > in proximity to each other.
> >
> > > My molecule is somewaht
> > > similar to plastocyanin as discussed in AMBER tutorial 4 instead of
> > copper
> > > I have a K+ ion (non -bonded interaction with monomer) , I have non
> > bonded
> > > parameters in my frcmod file while I construct the solvated box .
> >
> > By the way, the standard ion vdw parameters are derived for interactions
> > with the water model, not the Amber residues and others parameterized
> > the same way. I've never seen anyone else stress this fact, but
> > here is a paper where it is investigated at length:
> >
> > Ion-Induced Stabilization of the G-DNA Quadruplex: Free Energy
> > Perturbation Studies. W.S. Ross and C.C. Hardin
> > Journal of the American Chemical Society 116, 6070 (1994).
> >
> > > Is my error related to this intitial structure , I require to have a non
> > > bonded ion, how can I work with this in the input files while running
> > MD.
> >
> > A non-bonded ion is fine, see many studies of this. The first step of
> > triage is to rule out equilibration errors. Please re-read the below:
> >
> > > > equilibration at 50K at constant
> > > > volume and then I turn on constant pressure equilibration
> > >
> > > Did you jump the temperature or change the time step between
> > > the two? I suggest pasting in all your mdin files to the response.
> > >
> > > Bill
> > >
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Sun Dec 04 2011 - 14:30:03 PST
