[gmx-users] Re: pbc-replica exchange-trjconv

[servaas michielssens]

Tsjerk and Justin thanks for your suggestions.

Justin, the combination you suggested did not work, I tried many things 
without succes here.
I got the following error using whole and mol. (after doing nojump)
There were 8 inconsistent shifts. Check your topology
There were 8 inconsistent shifts. Check your topology
Will stop reporting inconsistent shifts

Tsjerk you are right I should explain the names:
trjconv -f gromacs.trr -o cluster.trr -n lig_prot.ndx -pbc cluster -s 
top140.tpr

gromacs.trr = trajectory generated by replica exchange
cluster.trr = output name
lig_prot.ndx=index file of with a group for ligand and protein
top140.tpr=tpr file generated using the starting configuration of 
gromacs.trr

You were right about the fit last time, but the error stays the same if I 
don't do the fit first...
Could the it be the fact that it is a replica exchange simulation that 
causes the problem, this makes that there are jumps in the trajectory of 
course...

kind regards,

servaas



> servaas michielssens wrote:
>> Dear gromacs users,
>>
>> I have a problem that is already discussed a lot on the mailing list,
>> in a protein-ligand simulation the ligad jumps out of the box. The
>> trajectory is generated by replica exchange simulation. So I used
>> trjconv with the option cluster:
>>
>> trjconv -f fit.trr -o cluster.trr -n lig_prot.ndx -pbc cluster -s
>> top140.tpr
>> The program seems to get stuck at frame 208, repeating the following
>> lines:
>> COM:    1.730     1.730     2.447  iter = 6214  Isq =   21.428
>> COM:    0.000     0.000     0.000  iter = 6215  Isq =   54.757
>> COM:    1.730     1.730     2.447  iter = 6208  Isq =   21.428
>> COM:    0.000     0.000     0.000  iter = 6209  Isq =   54.757
>> COM:    1.730     1.730     2.447  iter = 6210  Isq =   21.428
>> COM:    0.000     0.000     0.000  iter = 6211  Isq =   54.757
>> COM:    1.730     1.730     2.447  iter = 6212  Isq =   21.428
>>
>> if I use this command:
>> trjconv -f fit.trr -o cluster.trr -n lig_prot.ndx -pbc cluster
>>
>> The program runs but the ligands is still out of the box, the option
>> -s seems to be necessary, but not working in my case.
>>
>> I also tried the -nojump option and after this the whole option but in
>> visualistion (with VMD) I got strange bonds...
>>
>
> I've found that sometimes several iterations of trjconv are necessary to
> get such things fixed up.  Something like -pbc nojump, followed by
> 'whole,' and 'mol' or 'res' after that might do the trick.  It's a bit
> of trial and error, and if anyone else out there has a better method,
> I'd love to hear it, too :-)
>
> -Justin
>
>> An often reported problem was that the the structure in the tpr file
>> is not close enough to the starting structure in the trajectory, I
>> tried it by making a tpr file with a the strating structure of the
>> trajectory (in this structure the ligand is in the active site if I
>> look at the structure). But this did not help.
>>
>>
>> What I am doing wrong here?
>>
>> thanks in advance for your help!
>>
>> kind regards,
>>
>> servaas
>> _______________________________________________


> Hi Servaas,
>
>> trjconv -f fit.trr -o cluster.trr -n lig_prot.ndx -pbc cluster -s 
>> top140.tpr
>
> I think there's quite a bit of reason to start calling you names here :)
> Assumedly, fit.trr means that it results from fitting the trajectory
> to a reference?
> So what does this do with your PBC?
>
>> I also tried the -nojump option and after this the whole option but in
>> visualistion (with VMD) I got strange bonds...
>
> Right, so strange bonds should be quite an indication that there's
> something weird with your shifts.
>
>> An often reported problem was that the the structure in the tpr file is 
>> not
>> close enough to the starting structure in the trajectory, I tried it by
>> making a tpr file with a the strating structure of the trajectory (in 
>> this
>> structure the ligand is in the active site if I look at the structure). 
>> But
>> this did not help.
>
> Another often, or at least several times, mentioned problem relates to
> the fact that fitting a structure reorients the molecule, but leaves
> the PBC as it is, causing a mismatch between the system and the PBC.
> Therefore one can not execute any PBC related operations (nojump,
> cluster, whole), after the trajectory has been fitted.
>
> No hard feelings ;)
>
> Cheers,
>
> Tsjerk

>