[gmx-users] RAM usage of gmx msd

[Martin]

Hi,

thanks a lot for your answer. It contained exactly the information that 
I needed.
Here is a summary of my issue and how I resolved it:

Problem:

I wanted to calculate lateral diffusion of lipids in a cell membrane 
using a key atom for each lipid. But gmx msd needs too much RAM.

Solution:

I first created a smaller trajectory file using:
gmx trjconv -f mdrun.xtc -n key_atoms.ndx -o key_atoms.xtc
Then I adjusted the tpm file:
gmx convert-tpr -s mdrun.tpr -n key.ndx -o key_atoms.tpr
I have now limited the trajectory to only the atoms that I actually need.
But now the original index file no longer fits.
So I created a new one using:
gmx make_ndx -f key_atoms.tpr -o key_atoms.ndx

Once that was done I was able to calculate gmx msd with barely any RAM 
usage at all.

Thanks again for all your support.
Best regards
Martin Kern

Am 18.09.19 um 11:52 schrieb Mark Abraham:
> Hi,
>
> It's likely that using the same index file is the problem. The numbers it
> contains are interpreted relative to the tpr file, so if you make a subset
> of the tpr file, then it's on you to understand whether the necessary
> indices have changed or not.
>
> Mark
>
> On Tue, 17 Sep 2019 at 14:58, Martin Kern <martin.kern at uni-konstanz.de>
> wrote:
>
>> Hi John,
>>
>> thanks for your answer. The mdrun_key_atoms.xtc FILE is only around 500
>> MB in size. But the issue here wasn't RAM usage but a segmentation
>> fault. gmx msd requires a tpr file and I guess that my tpr and xtc files
>> didn't fit together. Skipping frames might work without the need of a
>> modified tpr file. I'll try that tomorrow.
>>
>> Am 17.09.2019 um 12:09 schrieb John Whittaker:
>>> Hi Martin,
>>>
>>>
>>>> Hello everyone.
>>>>
>>>> I simulated a cell membrane and would like to calculate lateral
>>>> diffusion of lipids. I tried this using the gmx msd command.
>>>> Unfortunately this uses enormous amounts of RAM. The process runs
>>>> without error until it is killed by the operating system. No output file
>>>> is created at that time.
>>>>
>>>> The membrane contains around 400 lipids and I simulated for 1100ns which
>>>> is 220000 frames. The total size of the xtc file is around 150 GB. I use
>>>> GROMACS 2016.4 with GPU support. The command I used was:
>>>> gmx msd -s mdrun.tpr -f mdrun.xtc -n key_atoms.ndx -lateral z -o
>>>> lateraal_diffusion.xvg
>>>>
>>>> I found an old email that also mentions the high RAM usage of gmx msd
>>>> but it didn't get a reply.
>>>>
>> https://mailman-1.sys.kth.se/pipermail/gromacs.org_gmx-users/2018-January/118014.html
>>>> I also tried reducing the RAM usage by creating a trajectory that only
>>>> includes the key atoms. This attempt resulted in a segmentation fault.
>>>> Here is what I tried:
>>>> gmx trjconv -f mdrun.xtc -n key_atoms.ndx -o mdrun_key_atoms.xtc
>>>> gmx convert-tpr -f mdrun.tpr -n key_atoms.ndx -o mdrun_key_atoms.tpr
>>>> gmx msd -s mdrun_key_atoms.tpr -f mdrun_key_atoms.xtc -n key_atoms.ndx
>>>> -lateral z -o lateraal_diffusion.xvg
>>> How big is the "mdrun_key_atoms.xtc" trajectory? It's possible that this
>>> file is still too large for the amount of RAM available on your machine.
>>>
>>> Are you able to break this trajectory down into smaller chunks that are
>>> more manageable? You could also use the -skip or -dt options of gmx
>>> trjconv in order to write only every "n" frames.
>>>
>>> Hope this helps,
>>>
>>> John
>>>
>>>
>>>> I'm grateful for any suggestion.
>>>>
>>>> Best regards
>>>> Martin Kern
>>>>
>>>> --
>>>> Gromacs Users mailing list
>>>>
>>>> * Please search the archive at
>>>> http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before
>>>> posting!
>>>>
>>>> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>>>>
>>>> * For (un)subscribe requests visit
>>>> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
>> send
>>>> a mail to gmx-users-request at gromacs.org.
>>>>
>> --
>> Gromacs Users mailing list
>>
>> * Please search the archive at
>> http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before
>> posting!
>>
>> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>>
>> * For (un)subscribe requests visit
>> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
>> send a mail to gmx-users-request at gromacs.org.
>>