[gmx-users] Extending an mdrun with trr >2GB

[Martina Bertsch, PhD]

Hi, David.

I am not sure what you mean. On our cluster, my home directory with gmx 
data and the trr > 2.2 GB is already NFS mounted on all nodes. What 
would I do next?

As I mentioned earlier, the command:

tpbconv -s full.tpr -f full.trr -o full_extend_1.tpr -extend 10000


does not work, because the full.trr "cannot be found". Peter Zoon 
suggested compressing trr into xtc format, but that will not work, 
because you need velocity information to continue or extend a run.

Would it be terribly wrong if I simply started a new 10-ns simulation 
with the final coordinates from the first trr:

grompp -f full.mdp -c MT1_popc_postfull.gro -n index.ndx -p MT1_popc.top -o full_extend.tpr

i.e., if I disregarded the velocities?

What I am trying to do is collect the full MD data for my receptor in 
the POPC bilayer.

Thank you for your clarifications.
Martina


David wrote:

>On Sat, 2004-03-27 at 10:10, Martina Bertsch, PhD wrote:
>
>>If I understand correctly, trjconv converts the trr file to xtc format, 
>>which does *not* contain velocities or forces and thus saves space. 
>>Manuals specifically instruct us not to remove the trr file if a run has 
>>to be restarted.
>>
>
>You need the trr file. This is a linux problem that has been discussed
>many times on the list. Apparently we need a FAQ entry. A possible
>solution is to NFS mount the disk your trajectory is on and do it over
>NFS. With a decently modern Linux it should not be a problem (we run
>Redhat 8 and higher):
>
>[rembrandt:octanol/XTC] % ls -l simE2.xtc
>-rw-r--r--    1 spoel    xray     16029520596 Nov 15 21:15 simE2.xtc
>
>
>
>>I don't think I would get a true continuation run without velocities. Am 
>>I correct?
>>
>>Martina 
>>
>>Peter Zoon wrote:
>>
>>>Hi!
>>>with trjconv you can make full.trr smaller and thus overcome the 
>>>problem of file> 2GB
>>>Then you can use this newly made file to restart, with tpbconv.
>>>
>>>Cheers
>>>Peter
>>>
>>>At 09:18 3/27/2004, you wrote:
>>>
>>>>Greetings.
>>>>
>>>>Our administrator had to install upgrades on the cluster, so I had to 
>>>>stop my mdrun of a protein in a POPC bilayer, which was running 
>>>>smoothly up until now (for 10 ns). To stop the run, I used the kill 
>>>>-USR1 signal. The signal produced a gro file, named 
>>>>"MT1_popc_postfull.gro", and the megaflops accounting report at the 
>>>>end of the log file, as if it were the end of the run. However, the 
>>>>run would have been continued for another 10 ns, if there had not 
>>>>been any interruption.
>>>>
>>>>When the upgrades were completed, I attempted to restart the run with 
>>>>the following command:
>>>>
>>>>tpbconv -s full.tpr -f full.trr -o full_extend_1.tpr -extend 10000
>>>>
>>>>I also tried the following:
>>>>
>>>>grompp -f full.mdp -t full.trr -c MT1_popc_postfull.gro -n index.ndx 
>>>>-p MT1_popc.top -o full_extend.tpr
>>>>
>>>>But in both cases, I received an error message:
>>>>
>>>>File full.trr not found.
>>>>
>>>>Of course, the file is actually present and larger than 2.2 GB.
>>>>
>>>>How do I continue/extend my mdrun? I could start a new 10 ns 
>>>>simulation with the final coordinates from MT1_popc_postfull.gro, but 
>>>>it would not be a true continuation run without the velocities from 
>>>>the first part of the run, since I could not read the trr file.
>>>>
>>>>Also, will I be able to analyze the trr trajectory that is larger 
>>>>than 2 GB with g_rms, g_hbond etc, or will I have to convert it to 
>>>>the xtc format?
>>>>
>>>>Thank you.
>>>>
>>>>Martina
>>>>
>>>>_______________________________________________
>>>>gmx-users mailing list
>>>>gmx-users at gromacs.org
>>>>http://www.gromacs.org/mailman/listinfo/gmx-users
>>>>Please don't post (un)subscribe requests to the list. Use the www 
>>>>interface or send it to gmx-users-request at gromacs.org.
>>>>
>>>
>>>
>>>_______________________________________________
>>>gmx-users mailing list
>>>gmx-users at gromacs.org
>>>http://www.gromacs.org/mailman/listinfo/gmx-users
>>>Please don't post (un)subscribe requests to the list. Use the www 
>>>interface or send it to gmx-users-request at gromacs.org.
>>>
>>>
>>
>>_______________________________________________
>>gmx-users mailing list
>>gmx-users at gromacs.org
>>http://www.gromacs.org/mailman/listinfo/gmx-users
>>Please don't post (un)subscribe requests to the list. Use the 
>>www interface or send it to gmx-users-request at gromacs.org.
>>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20040327/a3934b5d/attachment.html>