[gmx-users] monoclinic systems and pbc

[Tsjerk Wassenaar]

Hi Nick (or Claus?, I'm confused),

>  For the monoclinic axes I enter
>
>  ax = 4*0.663, by = 4*2.078,          cz =
> 4*0.665*sin(a^c)
>  ay = 0,       az = 0,                bx = 0,
>  bz = 0,       cx = 4*0.665*cos(a^c), cy = 0
>
>  Question 1: Am I passing correctly my system to
> gromacs, or should I use directly the monoclinic
>  coordinates (I guess no)?

Basically, this is correct. You could feed a .pdb file with the
correct CRYST1 record through editconf, in which case you'd get the
corresponding .gro definition. However, Gromacs wants to have the
largest vector first, then the second largest vector, and finally the
shortest vector (see periodic boundary conditions, chapter 3 in the
manual). Thus, you will have to rotate your system to meet these
conditions. This will not change anything else in your system.

>
>  I have found out from the gromacs community
>  that the pbc in the mdp file in the case of a crystal
> should be better put as pbc=full.
>
>  Question 2: What is the difference between pbc=full
> and pbc=xyz.
>

pbc=full is indeed for use with (infinite) crystal lattices. In your
case, you have separate chains, and you should use pbc=xyz, as far as
I know.

>
>   The output gro file preserves the form of the
> monoclinic system (something that might be
>   against the philosophy of gromacs since it is stated
> in the manual that the simulation
>   is held in a brick shaped box).

Read the chapter in the manual on periodic boundary conditions carefully.

>   I can see again a monoclinic system, yet shorter in
> the direction of the inclined axis,
>   i.e., the initial chains have been folded inside the
> shorter monoclinic cell. They also
>   seem to be cut in one point and once again I see the
> message of the 64 inconsistent shifts. That is,
>   I do not recover the original input supercell.

I think this is due to the fact that your box does not meet the
gromacs conditions.

>
>   Question 3: Both with pbc=(xyz | full) I face
> problems, i.e.,
>   (inconsistent shifts | cannot recover original cell,
> broken chains).
>   Is it sth wrong in the way I load my system to
> gromacs, or should I check my topology file
>   for errors, which I have not yet managed to locate.

See above...

>
>   Question 4: Should I stick with pbc=full?
>               What is the meaning of inconcistent
> shifts?
>

I think you should pbc=xyz.., regarding the inconsisten shifts, I'll
pass that one :)

>   Question 5: Should I change the way I use trjconv?
>               I guess that with the use of trjconv,
> from the output cubic cell of gromacs
>               I should be able to recover the initial
> monoclinic supercell.
>

No you don't need to change the way you used trjconv, but you have to
be sure that the initial setup of your simulation cell is correct.

>
>   Thank you very much,
>   Nick
>
You're most welcome ;)

Cheers,

Tsjerk

-- 
Tsjerk A. Wassenaar, Ph.D.
Junior UD (post-doc)
Biomolecular NMR, Bijvoet Center
Utrecht University
Padualaan 8
3584 CH Utrecht
The Netherlands
P: +31-30-2539931
F: +31-30-2537623