[gmx-users] weird output from simultions of a protein between two sheets

[Justin Lemkul]

On 4/17/20 9:16 AM, Zuzana Benkova wrote:
> Dear Justin,
>
> thank you for your prompt response. I will answer your questions step by step.
>
> Is this just in the final coordinate file? Are coordinates in the
> trajectory correct when visualized?
>
> No, I converted all compressed trajectories to gro format and this outputs are identical in all trajectories. When I visualize the trajectory, I can see it as well. The atoms are placed in apices of boxes (1x1x1).

This is clearly a bug. Please report it on 
https://gitlab.com/groups/gromacs/-/issues and include a .tpr file.

Freezing is, in any case, completely artificial and results in 
collisions that do not conserve energy. Try a stiff position restraint 
instead.

>
> Are you sure you don't want to remove artificial contributions to COM
> motion? This is unusual for a periodic system.
>
> I am not sure. If I leave the Linear option does the correction apply to all system even though I freeze the coordinates of graphene sheets? Anyway, I have just performed the same simulation with comm-mode = Linear and the problem persists.

The use of COM motion removal isn't directly related to your issue, but 
you should still treat it properly.

-Justin

> Saving output every step is unnecessary (because you will be skewing
> your statistics with completely correlated frames) and also will
> severely degrade the performance of your simulation.
>
> Thank you for this notification. I reduce the frequency when I do equilibration and production run. I do block averages. I have saved the data every step because I tried to trace the problem.
>
> Greetings
>
> Zuzana
>   
>
>
>
> ----- Original Message -----
> From: "Justin Lemkul" <jalemkul at vt.edu>
> To: "gmx-users" <gmx-users at gromacs.org>
> Sent: Friday, April 17, 2020 1:48:28 PM
> Subject: Re: [gmx-users] weird output from simultions of a protein between two sheets
>
> On 4/17/20 7:14 AM, Zuzana Benkova wrote:
>> Dear Gromacs users,
>>
>> I want to simulate a protein between two graphene sheets separated by 21 nm. The system is solvated and placed in the cell of dimensions 17.09360  17.18200  25.00000.
>> The energy minimzation of this system with position restrain applied to protein and frozen carbon atoms ran OK.
>> I continued with short simulations in an NVT ensemble with position restrains applied to protein. My output coordinates for all atoms of the systems were integers. Small fragment of a selected frame of the output is as follows
>>
>>       1GRA     C1    1    0    0    0
>>       1GRA     C2    2    0    0    0
>>       1GRA     C3    3    0    0    0
>>       1GRA     C4    4    0    0    0
>>       2GRA     C1    5    0    0    0
>>       2GRA     C2    6    0    1    0
>>       2GRA     C3    7    0    1    0
>>       2GRA     C4    8    0    1    0
>>       3GRA     C1    9    0    1    0
>>       3GRA     C2   10    0    1    0
>>       3GRA     C3   11    0    1    0
>>       3GRA     C4   12    0    1    0
> Is this just in the final coordinate file? Are coordinates in the
> trajectory correct when visualized?
>
>> my mdp parameters are as follows
>>
>> integrator               = md
>> tinit                    = 0
>> dt                       = 0.001
>> nsteps                   = 100
>> comm-mode                = None
> Are you sure you don't want to remove artificial contributions to COM
> motion? This is unusual for a periodic system.
>
>> nstcomm                  = 0
>> comm-grps                =
>> freezegrps               = Gra1 Gra2
>> freezedim                = Y Y Y Y Y Y
>> energygrps               = Gra1 Gra2 Protein SOL NA
>> ;energygrp-excl           = Gra1 Gra1 Gra2 Gra2 Gra1 Gra2
>> nstxout                  = 0
>> nstvout                  = 0
>> nstfout                  = 0
>> nstlog                   = 1
>> nstcalcenergy            = 1
>> nstenergy                = 1
>> nstxout-compressed       = 1
>> compressed-x-precision   = 1
> Saving output every step is unnecessary (because you will be skewing
> your statistics with completely correlated frames) and also will
> severely degrade the performance of your simulation.
>
> -Justin
>
>> pbc                      = xyz
>> periodic_molecules       = yes
>> cutoff-scheme            = Verlet
>> nstlist                  = 10
>> ns_type                  = grid
>> verlet-buffer-tolerance  = 0.005
>> rlist                    = 1.4   ;ignored with Verlet
>> coulombtype              = PME
>> coulomb-modifier         = None
>> rcoulomb-switch          = 1.1
>> rcoulomb                 = 1.3
>> vdw-type                 = Cut-off
>> vdw-modifier             = Force-switch
>> rvdw_switch              = 1.0
>> rvdw                     = 1.3
>> DispCorr                 = EnerPres
>> fourierspacing           = 0.12
>> pme-order                = 4
>> ewald-rtol               = 1e-05
>> ewald-geometry           = 3d
>> Tcoupl                   = V-rescale
>> nsttcouple               = -1
>> tc-grps                  = Protein  SOL_NA  Gra1_Gra2
>> tau-t                    = 0.1  0.1  0.1
>> ref-t                    = 310  310  310
>> Pcoupl                   = no
>>
>>
>> I have to note that I got the same output with no position restraints on protein.
>>
>> Do you have any idea what the problem is?
>>
>> Thank you in advance.
>>
>> Greetings
>>
>> Zuzana
>>
>>
>>
>>
>>

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

Justin A. Lemkul, Ph.D.
Assistant Professor
Office: 301 Fralin Hall
Lab: 303 Engel Hall

Virginia Tech Department of Biochemistry
340 West Campus Dr.
Blacksburg, VA 24061

jalemkul at vt.edu | (540) 231-3129
http://www.thelemkullab.com

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