On Fri, May 9, 2014 at 2:52 AM, Valentina Romano
<valentina.romano.unibas.ch> wrote:
> Is there a way to set in a single input file all parameters for minimizations in which restraints are gradually relax from e.g. 2.o to 0.0? (or should I have an input file for each minimization?)
I don't think you can vary positional restraints the way you can e.g.
distance restraints. My recommendation anyway is to use a separate
input file for each minimization - this way you can monitor each phase
for problems.
-Dan
>
> Thanks a lot.
> Valentina
> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> Valentina Romano | PhD Student | Biozentrum, University of Basel & SIB Swiss Institute of Bioinformatics
> Klingelbergstrasse 61 | CH-4056 Basel |
>
> Phone: +41 61 267 15 80
>
>
> ________________________________________
> From: Daniel Roe [daniel.r.roe.gmail.com]
> Sent: Thursday, May 08, 2014 7:10 PM
> To: AMBER Mailing List
> Subject: Re: [AMBER] Analysis of minimization stage
>
> On Thu, May 8, 2014 at 10:32 AM, Valentina Romano
> <valentina.romano.unibas.ch> wrote:
>> cut=12,
>
> This cutoff might be unnecessarily large for PME. You can probably use
> 10 or even 8.
>
>> tempi=0.0,
>> temp0=300.0,
>
> I am not a huge fan of heating up from 0 K (unless that is what you
> specifically need to study for some reason). I feel that assigning
> random velocities is much better, otherwise you're really asking a lot
> from the thermostat - especially here where you're essentially asking
> for an instant jump from 0 to 300 K. I have never had a problem with
> assigning random velocities at the temperature of interest (in this
> case you would want tempi=300). Make sure you also set ig=-1 (this is
> good advice for MD in general unless you need to specifically
> reproduce some result).
>
>> ntt=3,
>> gamma_ln=1.0,
>
> My personal preference for equilibration is to use the Berendsen
> thermostat with tight coupling (ntt=1, tautp=0.5), but Langevin may be
> OK as well (although I would increase your gamma_ln to at least 5, and
> maybe higher).
>
>> nstlim=50000, dt=0.002,
>
> This is probably longer than you need for this stage, and as was
> previously suggested definitely use a 1 fs timestep.
>
>> ntpr=100, ntwx=100, ntwr=1000
>> restraint_wt = 1.0,
>> restraintmask=':1-247.CA,C,O,N'
>
> I like a stronger restraint initially (5 kcal/mol Ang^2), which can be
> gradually relaxed in subsequent phases.
>
> You can follow this phase with a few minimizations in which you
> gradually relax the restraints (e.g. 2.0 to 0.1 to nothing). Then you
> can switch to NPT with restraints on again (at this point 1.0 kcal/mol
> Ang^2 is probably fine); use several short (5-10 ps) runs in which you
> again gradually relax the restraints. Keep a 1 fs timestep throughout.
> When you finally have all restraints off be sure to monitor your
> system density - when that stops changing significantly your system
> should be in decent shape.
>
> Hope this helps,
>
> -Dan
>
>>
>>
>> Do you have specif suggestion about it?
>>
>> -Valentina
>>
>>
>> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
>> Valentina Romano | PhD Student | Biozentrum, University of Basel & SIB Swiss Institute of Bioinformatics
>> Klingelbergstrasse 61 | CH-4056 Basel |
>>
>> Phone: +41 61 267 15 80
>>
>>
>> ________________________________________
>> From: Daniel Roe [daniel.r.roe.gmail.com]
>> Sent: Thursday, May 08, 2014 6:21 PM
>> To: AMBER Mailing List
>> Subject: Re: [AMBER] Analysis of minimization stage
>>
>> Hi,
>>
>> On Thu, May 8, 2014 at 10:04 AM, Valentina Romano
>> <valentina.romano.unibas.ch> wrote:
>>> First of all, I minimized the system in 2 steps.
>>> 1) only solute restrained
>>> 2) all atoms free
>>> The input file for the 2nd step is:
>>>
>>> &cntrl
>>> imin=1,
>>> maxcyc=10000,
>>> ncyc=2500,
>>> ntb=1,
>>> igb=0,
>>> cut=12
>>> /
>>
>> You don't say what input you used for the MD stage, but explicit
>> solvent simulations in general require more careful minimization than
>> implicit solvent sims. For example, what I usually do is after an
>> initial minimization with relatively strong positional restraints on
>> solute heavy atoms, my second stage will be a short (15 ps) NVT MD
>> simulation with the same restraints using a 1 fs timestep to let the
>> water relax, followed by more minimization with gradually decreasing
>> restraints. I then switch to a series of short NPT simulations with
>> gradually decreasing restraints on solute to allow the density to
>> start to equilibrate. If you post your actual MD input we could make
>> more specific recommendations.
>>
>>> I think the error is related to the minimized structure I am using as input.
>>> Is there a way to check if the minimization went well (e.g.: an easy way to plot the total energy vs time steps)?
>>> How to check if it necessary minimize for a longer or shorter time?
>>
>> I think that MdoutAnalyzer.py will let you plot energy terms from
>> MDOUT output relatively easily. Alternatively you could read MDOUT
>> data in with cpptraj (i.e. 'readdata <mdout>'), then write whatever
>> terms you want to the file format of your choice with 'writedata'. In
>> general you want to pay attention to your gradient - if it's not
>> relatively flat for your last steps you probably could do with
>> additional minimization. Alternatively you may want to try using the
>> XMIN minimizer in sander (ntmin = 3).
>>
>> Good luck,
>>
>> -Dan
>>
>>>
>>> The last step of the minimization is the following:
>>>
>>> NSTEP ENERGY RMS GMAX NAME NUMBER
>>> 10000 -1.0588E+05 6.6968E-01 7.0286E+01 N9 3834
>>>
>>> BOND = 7509.5321 ANGLE = 683.5019 DIHED = 2272.1178
>>> VDWAALS = 17150.2089 EEL = -143037.0880 HBOND = 0.0000
>>> 1-4 VDW = 785.8019 1-4 EEL = 8760.7170 RESTRAINT = 0.0000
>>>
>>> While the enrgy at the step 0 of the MD is:
>>> NSTEP = 0 TIME(PS) = 0.000 TEMP(K) = 0.00 PRESS = 0.0
>>> Etot = -113244.2583 EKtot = 0.0000 EPtot = -113244.2583
>>>
>>> Could the difference in the energy values had influenced the error I got? If yes, I can solve it?
>>>
>>> Thank you
>>> Valentina
>>> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
>>> Valentina Romano | PhD Student | Biozentrum, University of Basel & SIB Swiss Institute of Bioinformatics
>>> Klingelbergstrasse 61 | CH-4056 Basel |
>>>
>>> Phone: +41 61 267 15 80
>>>
>>>
>>>
>>> _______________________________________________
>>> AMBER mailing list
>>> AMBER.ambermd.org
>>> http://lists.ambermd.org/mailman/listinfo/amber
>>
>>
>>
>> --
>> -------------------------
>> Daniel R. Roe, PhD
>> Department of Medicinal Chemistry
>> University of Utah
>> 30 South 2000 East, Room 201
>> Salt Lake City, UT 84112-5820
>> http://home.chpc.utah.edu/~cheatham/
>> (801) 587-9652
>> (801) 585-6208 (Fax)
>>
>> _______________________________________________
>> AMBER mailing list
>> AMBER.ambermd.org
>> http://lists.ambermd.org/mailman/listinfo/amber
>>
>> _______________________________________________
>> AMBER mailing list
>> AMBER.ambermd.org
>> http://lists.ambermd.org/mailman/listinfo/amber
>
>
>
> --
> -------------------------
> Daniel R. Roe, PhD
> Department of Medicinal Chemistry
> University of Utah
> 30 South 2000 East, Room 201
> Salt Lake City, UT 84112-5820
> http://home.chpc.utah.edu/~cheatham/
> (801) 587-9652
> (801) 585-6208 (Fax)
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
--
-------------------------
Daniel R. Roe, PhD
Department of Medicinal Chemistry
University of Utah
30 South 2000 East, Room 201
Salt Lake City, UT 84112-5820
http://home.chpc.utah.edu/~cheatham/
(801) 587-9652
(801) 585-6208 (Fax)
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Fri May 09 2014 - 07:30:03 PDT
