VMD-L Mailing List

From: Gumbart, JC (gumbart_at_physics.gatech.edu)
Date: Sun Oct 24 2021 - 23:26:45 CDT

I have personally found Jerome Henin’s qwrap a little more intuitive in its behavior than pbctools. You might want to give a try and see if you have better luck.
https://urldefense.com/v3/__https://github.com/jhenin/qwrap__;!!DZ3fjg!qHZjQvzIIdt26MYmoUZmWbM8ozfgBpiIGEz7RmUc0yi20Fx6nw7cghb9PnA31ocPTw$

Best,
JC

On Oct 24, 2021, at 2:47 PM, Ryan Woltz <rlwoltz_at_ucdavis.edu<mailto:rlwoltz_at_ucdavis.edu>> wrote:

Thanks everyone for responding,

           It would be a tragedy if I have to re-run these. I have 9 systems with 1000ns of production time = 1500 hours (50+ days) on a V100 GPU so if there is a way to fix this. Not all systems have peptide jumping, maybe 6 out of the nine. There are 3 systems run in triplicate some jump some don't but the channel is behaving the same in the triplicates regardless of peptide jumping. The jumping really just starts anytime after 600ns if it does happen. I can correct for peptide jumping and get the protein in a box using the pbc join/wrap/unwrap commands.

    Membranes can behave unexpectedly, at times, so you may need to use restrains to keep those in place. You can apply position restraints in z direction so that your molecules are free to move in x and y direction but do not move out of the membrane.

      I'm using NAMD3 for the production which is about 3X times faster than NAMD2.14 for my system. NAMD3 however doesn't take restraints in. The membrane itself looks fine and similar to a membrane only system. layers aren't splitting or bubbling and the jumping is in the X-Y axis, I don't have any Z-dimension movement so given this would you still suggest restraints in the Z-axis? Is there a way to provide some screen shots to the community as it might help with troubleshooting? I think I understand why it is suggested to re-run, but I'm not seeing anything abnormal with the protein or membrane, besides the occasional lipid jumping and peptide jumping which I can fix with the pbc commands. Although as stated in the first email when I run the pbc commands the water is dispersed throughout many boxes, but loads in a single box so I think this has something to do with the selection I make during the pbc commands? For easy reference these are the wrapping commands I use: If I don't use -bondlist in my first command it doesn't disperse the water but the peptides aren't fixed all the way.

  *
pbc join fragment –bondlist –now
  *
pbc wrap –centersel protein –center com –compound fragment –all
  *
pbc unwrap

Thanks everyone for the suggestions,

Ryan

On Sat, Oct 23, 2021 at 6:37 AM Raman Preet Singh <ramanpreetsingh_at_hotmail.com<mailto:ramanpreetsingh_at_hotmail.com>> wrote:
Hello Ryan,

Do you see the jumps happening after equilibration or production run?

I have performed some membrane simulations using Gromacs, not NAMD, and, in general, the water molecules and other species could appear wierd during NVT equilibration steps, especially the water molecules. A short equilibration with NPT brings it back to "normal".

Membranes can behave unexpectedly, at times, so you may need to use restrains to keep those in place. You can apply position restraints in z direction so that your molecules are free to move in x and y direction but do not move out of the membrane.

Another problem can be the barostat that you are using. CHARMM-GUI settings usually work well but you never know where the usual settings could fail.

Regards,
Raman

Get Outlook for Android<https://urldefense.com/v3/__https://aka.ms/AAb9ysg__;!!DZ3fjg!vEqoXdRviPWuxWM0aPhZ1SCGvTuJW17xsUucwFPDyG4-pYFxEn_jCiSU32dPAGh1gg$>
________________________________
From: owner-vmd-l_at_ks.uiuc.edu<mailto:owner-vmd-l_at_ks.uiuc.edu> <owner-vmd-l_at_ks.uiuc.edu<mailto:owner-vmd-l_at_ks.uiuc.edu>> on behalf of Ashar Malik <asharjm_at_gmail.com<mailto:asharjm_at_gmail.com>>
Sent: Saturday, October 23, 2021 11:47:03 AM
To: Ryan Woltz <rlwoltz_at_ucdavis.edu<mailto:rlwoltz_at_ucdavis.edu>>
Cc: VMD Mailing List <vmd-l_at_ks.uiuc.edu<mailto:vmd-l_at_ks.uiuc.edu>>
Subject: Re: vmd-l: protein and lipid jumping

Hi,

Unfortunately, I cannot be of any help in this regard.
I have never encountered this problem before and if I did, I would actually run the MD simulation again.

As I mentioned earlier as well, it is my hunch that the behaviour you are seeing is because of periodic boundaries not being set up correctly.

I am not familiar with any steps that can fix your problem, but then again I am not an expert.

Perhaps you could wait here for more input from others who are more experienced. both to confirm the cause of the problem and if someone might have seen this behaviour they might suggest a solution.

Sorry for not being able to suggest a fix.

Regards,
Ashar

On Sat, Oct 23, 2021 at 2:01 PM Ryan Woltz <rlwoltz_at_ucdavis.edu<mailto:rlwoltz_at_ucdavis.edu>> wrote:
Hi Ashar,

       Yes, with your explanation I'm certain the initial boundaries is ~145. Sorry for the confusion but the naming is different with files coming out of Charmm-gui with the second email being initial config file in step1 of the setup. Charmm-gui goes through 5 stages of setup. The protein +membrane is setup in step3 which is where I enter the "145" number and it confirms the protein can fit in this size.

        The files that do the initial run of MD is a .inp (input) file and I found this line in here.

cellBasisVector1 $a 0.0 0.0; # vector to the next image
cellBasisVector2 $d $b 0.0;
cellBasisVector3 0.0 0.0 $c;
cellOrigin 0.0 0.0 $zcen; # the *center* of the cell

         So in the previous response the lines set a,b,c I think this is how these variables are being used. From file step5_input.str. These input values are again

set a 141.104004
 set b 141.104004
 set c 144.858064

so 141, 141, and 145. apologies for the misunderstanding of terms

       If the values are used as boundaries for your entire system (i.e., membrane + water) and the minmax result you provided earlier - then yes. Your periodic setup is smaller than your entire system which can explain why your "jumps" appear with the system instead of on edges.

        How do you recommend we proceed from here? I've been trying to wrap things using the protein as my selection and focusing on that but is there a way to wrap by size dimensions you calculated? (180,188,158) I know this will probably be very time consuming to do it this way, correct?

Thanks for your patience and help,

Ryan

On Fri, Oct 22, 2021 at 10:43 PM Ashar Malik <asharjm_at_gmail.com<mailto:asharjm_at_gmail.com>> wrote:
Hi Ryan

{"systype":"bilayer","dimensions":["145.078088","145.078088","144.858064","90.0","90.0","90.0"],"input":["namd"],"forcefield":{"type":"c36m","files":["default"],"custom":[]},"hmr":false}

Again not familiar with the contents of the string above and how each of the values is being used.
I am also not familiar with how you set up your system.

Taking your system as an example a typical workflow will look something like this

A membrane + protein system is constructed which is then minimized followed by solvation/ionization, minimized again and then the system goes into MD.

At each step when things are being moved or added your system size can vary.

Just before MD, you have to know the system size. The way I am familiar with MD is through the use of a config file which contains line-by-line a collection of variables that you can change the value of. Programs like NAMD will then read that list and the explicit values from the config file will replace system defaults. Some values have no defaults and must be provided.

e.g., see here https://www.ks.uiuc.edu/Training/Tutorials/namd/namd-tutorial-unix-html/node26.html

Within the description listed in the page above you will notice 4 lines.

# Periodic Boundary conditions
cellBasisVector1 dx 0. 0. ;# vector to the next image
cellBasisVector2 0. dy 0.
cellBasisVector3 0. 0 dz
cellOrigin 0. 0. 0. ;# the *center* of the cell

I have substituted the values to dx, dy, dz for generalization.
Your system size before MD (i.e. length along x, y and z axis) needs to be plugged into dx, dy and dz.
cell Origin will point to the centre of the box which can be 0, 0, 0 if it's at origin.

This now controls the boundaries across which the wrapping will take place. If your system is larger and the boundaries you choose are smaller then I suspect that you will get the behaviour you see. Disclaimer: I have never seen this in my work, but I suspect that the above might be the reason.

So I believe my boundaries are "145.078088","145.078088","144.858064" and compared to your calculations of 180 188 158 these do not match up, correct?

If the values are used as boundaries for your entire system (i.e., membrane + water) and the minmax result you provided earlier - then yes. Your periodic setup is smaller than your entire system which can explain why your "jumps" appear with the system instead of on edges.

For clarification, the charmm-gui of 135-145 is the approximate x,y,z dimensions you give Charmm-gui to create a system for you. So each system I have (9 of them) I used the input range from 135-145 angs depending on the system. For this system I'm using as an example I think I used 145 but there are others that are around 135. Sorry for the confusion, I think this is irrelevant information and whatever we fix in this system I can apply to the others.

Does this mean that 145 is your size in x, y and z?
Also - are you sure this (145) is the size of your entire system? Membrane + protein + waters?

Apologies for being new to this.

We all start from somewhere.

Hopefully the above explanation will clarify things.

Regards
Ashar