[gmx-users] g_select: selection file problem

[Teemu Murtola]

Hi,

On Thu, Apr 4, 2013 at 10:01 PM, Oleksandr Sushko
<o.sushko at eecs.qmul.ac.uk>wrote:

> I'm want to analyse stratified layers of water around protein,
> so I provide next selection file to g_select:

[clip]
> (shell2 is supposed to be a layer of whole water molecules from the layer
> 3 to 4 A)
> but it doesn't work correctly, moreover it even provides different results
> when just changing
> the order in 3rd line (shell2 = shell1 and shell), which shouldn't be the
> case.
> Interestingly, 1st and 2nd line separately works correctly.
>
> Even more interesting the next selection file works perfectly fine
> (regardless the order in 3rd line):
> [clip]
>

Thanks for reporting this, this is clearly a bug. I have filed
http://redmine.gromacs.org/issues/1216 with some explanation of what
triggers the bug.

(this also provides a layer between 3 and 4 A from protein, but some
> molecules are represented only by 1 or 2 atoms, which I want to avoid,
> that's why I'm using 'same residue as')
>

There is probably no way to get exactly your condition before the bug is
fixed, but if you are OK with a slightly altered condition, you can do, for
example, this (haven't tested, but should work):

shell = group "SOL" and res_com within 0.4 of group "protein";
shell1 = group "SOL" and not res_com within 0.3 of group "protein";
shell2 = shell and shell1;
shell2;

Best regards,
Teemu