[gmx-users] creating trajectory with ligand close to the protein

[David Osguthorpe]

On Sun, Feb 24, 2008 at 01:35:51PM +1100, Mark Abraham wrote:
> >
> >Ive been using trjconv but it seems to be failing for a case where the 
> >protein switches
> >image in the trajectory but the ligand does not.
> 
> Check out trjconv -h, and if none of the options works, please describe 
> why they're failing :-)
> 

Ive been using -pbc cluster with an index file of protein and ligand which has worked for other systems
(which is the closest Ive got to the algorithm I described previously although none of trjconvs options
or any other gromacs utility seem to implement the algorithm I described - if you could suggest which
utility/option is the closest to the above algorithm Id be very grateful)
but for a particular system it went into an infinite loop printing out
COM:    0.911     1.167     5.420  iter = 1
etc. (with iter going up forever) forever and a day.
Dumping coordinates before and after the step where this occurred and inspection by molecular graphics
showed that the ligand coordinates were similar but the protein coordinates were very different 
- which I assume means the protein shifted to a different periodic image - hence the comment.

David