[gmx-users] g_msd with input by trjconv -pbc nojump:a concern

Dommert Florian dommert at icp.uni-stuttgart.de
Mon Mar 19 15:05:43 CET 2012


perhaps this problem is related to bug 774:


which has been discussed quite often, recently. Somehow there is a
problem in the order of removing the PBC and jumps. As already
mentioned, I could solve the problem by providing a trajectory with
whole molecules to g_msd. An indexfile with seperate groups for every
molecule has to be created and finally every single molecule group has
to be analyzed, but WITHOUT the -mol flag. Afterwards the resulting MSDs
only need to be averaged.


On Mon, 2012-03-19 at 14:31 +0200, Ioannis Beis wrote: 
> Dear Gromacs users,
> I have been trying to calculate the lateral MSD of lipid molecules  
> within a bilayer. I have performed simulations in a rectangular box  
> with PBC. I have used trjconv with -pbc nojump. I compared results  
> between the initial trajectory and the one generated by trjconv for 6  
> different lipids. In 5 cases the calculated MSD was identical and only  
> in one there was difference (big difference). According to the  
> results, MSD loses linearity and exhibits large fluctuations after  
> some point. The above give rise to the following concern.
> As far as I understand, -pbc nojump checks if any molecule crosses the  
> box boundaries and if yes it brings back the broken part from the  
> symmetric side to the original place, keeping the molecule whole. This  
> way box information is lost, but each molecule is supposed to have  
> continuous trajectories. The true continuity of trajectories, however,  
> is only secured only if .xtc files carry all information of boundary  
> crosses based on the time step of the initial run in addition to all  
> coordinates for all times. Otherwise, the result is clearly dependent  
> on the frequency of saving and continuity is not guaranteed.
> E.g. in my case -I have small frequency of saving (100 ps) for long  
> simulations to avoid very large output files- one can't tell based  
> merely on the coordinates whether or not a lipid crossed boundaries  
> between two consecutive frames. Of course if the lipid went back and  
> forth then the result of the calculation would  not be affected but  
> only by statistical means; however if the lipid had an overall  
> translation across the box (not captured by -pbc nojump because the  
> molecule is whole in both frames), then the calculation is destroyed.
> If this is the case, it seems to me that g_msd might in practice give  
> correct results for proteins, but for smaller molecules the  
> reliability of the results depends sensitively on molecule  
> size-frequency of saving coordinates relationship and might give rise  
> to huge systematic errors.
> Is there a way to obtain a converted trajectory that guarantees  
> unconditional continuity, so that one makes sure that he obtains the  
> proper MSD for his molecules?
> Thank you in advance!
> Best regards,
> Ioannis

Florian Dommert
Dipl. - Phys.

Institute for Computational Physics
University Stuttgart

Pfaffenwaldring 27
70569 Stuttgart

EMail: dommert at icp.uni-stuttgart.de
Homepage: http://www.icp.uni-stuttgart.de/~icp/Florian_Dommert

Tel.: +49 - (0)711 - 68563613
Fax.: +49 - (0)711 - 68563658
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 198 bytes
Desc: This is a digitally signed message part
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20120319/bf0fbf11/attachment.sig>

More information about the gromacs.org_gmx-users mailing list