[gmx-users] Re: Umbrella sampling question

[Erik Marklund]

Hi,

Blindly defining the center of mass for a group of atoms is not possible in a periodic system such as a typical simulation box. You need some clue as to which periodic copy of every atom that is to be chosen. By providing pull_pbcatom0 you tell gromacs to, for every atom in grp0, use the periodic copy closest to the atom given by pull_pbcatom0. If you have large pullgroups this is necessary to define the inter-group distance in a way that makes sense. If you get different results depending on that setting you really need to figure out which atom is a good center for your calculations. The default behavior is to use the atom whose *index* is in the center of the group. If you for example have a dimeric protein this may correspond to the C-terminus of the first chain or the N-terminus of the second one, which in turn often doesn't coincide with the geometrical center of the group. I suggest you try yet another choice of pull_pbcatom0 that is also close to the center to see if that also give rise to a different distance. As mentioned, the choice of pull_pbcatom0 should not matter as long as the choice allows to figure out how to handle the periodicity.

Best,

Erik

15 nov 2012 kl. 19.56 skrev Gmx QA:

> Hi Chris
> 
> Seems my confusion was that I assumed that the distances in the
> profile.xvg-file should correspond to something I could measure with
> g_dist. Turns out it does not.
> Thank you for helping me sorting out this, I got it now :-)
> 
> About pull_pbcatom0 though. My box is >> 2*1.08 nm in all directions:
> $ tail -n 1 conf0.gro
>  12.45770  12.45770  17.99590
> 
> I am still not sure what pull_pbcatom0 does. You said it should not have
> any effect on my results, but changing it does result in a different
> initial distance reported by grompp.
> 
> In my initial attempts at this, I did not specify anything for
> pull_pbcatom0, but in the grompp output I get this
> 
> Pull group  natoms  pbc atom  distance at start     reference at t=0
>       0     21939     10970
>       1         1         0   2.083                 2.083
> Estimate for the relative computational load of the PME mesh part: 0.10
> This run will generate roughly 761 Mb of data
> 
> 
> Then, following the advice in the thread I referred to earlier, I set
> pull_pbcatom0 explicitly in the mdp-file to be an atom close to the COM of
> the Protein. Then I get from grompp
> 
> Pull group  natoms  pbc atom  distance at start     reference at t=0
>       0     21939      7058
>       1         1         0   1.808                 1.808
> Estimate for the relative computational load of the PME mesh part: 0.10
> This run will generate roughly 761 Mb of data
> 
> As you can see, the initial distance is different (2.083 vs 1.808), and
> 1.808 is the same as the distance reported by g_dist. Do you have any
> comments here as to why this is?
> 
> Thanks
> /PK
> 
> 
> What you reported is not what you did. It appears that grompp, gtraj, and
> g_dist report the same value.
> Please also report the value that you get from your pullx.xvg file that you
> get from mdrun, which I suspect
> will also be the same.
> 
> The difference that you report is actually between the first FRAME of your
> trajectory from g_dist
> and the first LINE of the file from the g_wham output. I see no reason to
> assume that the values in the
> output of g_wham must be time-ordered. Also, I have never used g_wham
> myself (I use an external program
> to do wham) and so I can not say if you are using it correctly.
> 
> My overall conclusion is that you need to investigate g_wham output not
> worry about a new run at this stage.
> 
> Regarding pull_pbcatom0, there is lots of information on the mailing list
> about this. It is a global atom number
> that defines the unit cell for selection of which periodic image of each
> molecule will be used for the pulling.
> If all of your box dimensions are >> 2*1.08 nm, then pull_pbcatom0 will not
> affect your results.
> 
> Chris.
> -- 
> gmx-users mailing list    gmx-users at gromacs.org
> http://lists.gromacs.org/mailman/listinfo/gmx-users
> * Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/Search before posting!
> * Please don't post (un)subscribe requests to the list. Use the 
> www interface or send it to gmx-users-request at gromacs.org.
> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists

-----------------------------------------------
Erik Marklund, PhD
Dept. of Cell and Molecular Biology, Uppsala University.
Husargatan 3, Box 596,    75124 Uppsala, Sweden
phone:    +46 18 471 6688        fax: +46 18 511 755
erikm at xray.bmc.uu.se
http://www2.icm.uu.se/molbio/elflab/index.html