[gmx-users] Inconsistent results in different clusters and cores

[Justin Lemkul]

On 5/12/13 1:53 PM, tarak karmakar wrote:
> Thanks,
>
> I have used CGENFF force field parameters for the ligand generated from
> PARMCHEM with 0 penalties. For protein I have used CHARMM36 force field.
> my npt.mdp file is as follows,
>
> ; 7.3.3 Run Control
> integrator              = md

Bug 1021 was only relevant with md-vv, so it is not your problem here.

> tinit                   = 0
> dt                      = 0.001
> nsteps                  = 5000000
> nstcomm                 = 1
> comm_grps               = system
> comm_mode               = linear
>
>
> ; 7.3.8 Output Control
> nstxout                 = 5000
> nstvout                 = 5000
> nstfout                 = 5000
> nstlog                  = 1000
> nstenergy               = 1000
> nstxtcout               = 1000
> xtc_precision           = 1000
> xtc_grps                = System
> energygrps              = lIG Protein Water
>
> ; 7.3.9 Neighbor Searching
> nstlist                 = 10
> ns_type                 = grid
> pbc                     = xyz
> rlist                   = 1.2
>
> ; 7.3.10 Electrostatics
> coulombtype             = PME
> rcoulomb                = 1.2
>
> ; 7.3.11 VdW
> vdwtype                 = cut-off
> rvdw                    = 1.2
> DispCorr                = EnerPres
>

Your short-range settings are incorrect for strict use of CHARMM.  You should set:

vdwtype = switch
rvdw-switch = 1.0
rlistlong = 1.4

Your other settings for rlist, rcoulomb, and rvdw are fine.

> ; 7.3.13 Ewald
> fourierspacing          = 0.12
> pme_order               = 4
> ewald_rtol              = 1e-5
>
> ; 7.3.14 Temperature Coupling
> tcoupl                  = nose-hoover
> tc_grps                 = system
> tau_t                   = 1.0
> ref_t                   = 300
>
> ; 7.3.15 Pressure Coupling
> pcoupl                  = parrinello-rahman
> pcoupltype              = isotropic
> tau_p                   = 1.0
> compressibility         = 4.5e-5
> ref_p                   = 1.0
>
> gen_vel                 = yes

In the absence of any restraints, initial velocity generation can produce 
incorrect dynamics.  This is why we use restraints.  Thus far, your observations 
simply seem consistent with random effects of improper nonbonded parameters 
and/or equilibration.

-Justin

> gen_temp                = 300
> gen_seed                = 8877691
>
> ; 7.3.18 Bonds
> constraints             = h-bonds
> constraint_algorithm    = LINCS
> continuation            = yes
> lincs_order             = 4
> lincs_warnangle         = 30
>
>
> On Sun, May 12, 2013 at 11:11 PM, Justin Lemkul <jalemkul at vt.edu> wrote:
>
>>
>>
>> On 5/12/13 1:34 PM, tarak karmakar wrote:
>>
>>> Thanks Justin for the Quick and Helpful reply.
>>>
>>>      Yes. If I am right, the chaotic behavior of the simulations is
>>> inherent
>>> and can be assessed statistically by generating several independent
>>> trajectories and analyzing their similar outcomes. But with the same
>>> '.mdp'
>>> file I am getting TOO much different results, and that's where I worry.
>>> I'll surely try with the recent version of gromacs. But, for now, can you
>>> give me a little more informations about problems (bugs) with the 4.5.5
>>> version, related to my context?
>>>
>>>
>> The proposed relationship to bug 1012 is unclear to me.  The issue there
>> was an incompatibility between an integrator and thermostat, with obvious
>> differences in thermodynamic output.  Assessing your system in this context
>> is not helpful.  If you want to assess whether different core counts or
>> hardware produce problems, then you need a very simple test case (like a
>> box of water) that shows significant differences in averaged observables.
>>   As I said before, maybe your ligand parameters are insufficiently accurate
>> (how did you generate them?) or .mdp settings are incorrect.  Without such
>> information, there is little point in trying to debug anything.
>>
>>
>> -Justin
>>
>> --
>> ==============================**==========
>>
>> Justin A. Lemkul, Ph.D.
>> Research Scientist
>> Department of Biochemistry
>> Virginia Tech
>> Blacksburg, VA
>> jalemkul[at]vt.edu | (540) 231-9080
>> http://www.bevanlab.biochem.**vt.edu/Pages/Personal/justin<http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin>
>>
>> ==============================**==========
>> --
>> gmx-users mailing list    gmx-users at gromacs.org
>> http://lists.gromacs.org/**mailman/listinfo/gmx-users<http://lists.gromacs.org/mailman/listinfo/gmx-users>
>> * Please search the archive at http://www.gromacs.org/**
>> Support/Mailing_Lists/Search<http://www.gromacs.org/Support/Mailing_Lists/Search>before posting!
>> * Please don't post (un)subscribe requests to the list. Use the www
>> interface or send it to gmx-users-request at gromacs.org.
>> * Can't post? Read http://www.gromacs.org/**Support/Mailing_Lists<http://www.gromacs.org/Support/Mailing_Lists>
>>

-- 
========================================

Justin A. Lemkul, Ph.D.
Research Scientist
Department of Biochemistry
Virginia Tech
Blacksburg, VA
jalemkul[at]vt.edu | (540) 231-9080
http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin

========================================