[gmx-users] Pull code

[Justin MacCallum]

Hi Michel,

The pull code is currently in a bit of disarray, which is entirely my
fault. I've been hacking bits and pieces of it as I need to for my work,
but I haven't had a chance to thoroughly test it. I've been planning to
do this for quite some time, but I just haven't had a chance. Also,
GROMACS has been undergoing a lot of development recently which makes it
harder to work on a feature like the pull code.

>   I am trying to make free energy measurements for the exchange of a 
> water molecule around a sodium ion via the Jarzynski method. The 
> principle is simple, steer a water towards the ion, and integrate the 
> forces. This system is maybe unusual for gromacs, but according to the 
> weird trajectories I get, it is doubtfull that the code work correctly 
> for any system...
> 
>   Here is a series of questions :
> 
> 1) When all groups are specified, why does mdrun ask for a group? The 
> group specified here doesen't seem to  influence the results.

This is a bug thats been fixed in the CVS version. It seems to be
harmless and everything should still work fine.

> 2) Using AFM, when a group crosses the box boundary, it is replaced b 
> its image on the other side of the box. But the steering potential stays 
> centered where it was, which results in an enormous force on the pulled 
> group, and fast crossing of the box.

We have used the AFM pulling options quite a lot, but never ran into
this problem due to the way our system was set up. At some point I would
like to fix this.

> 3) To prevent this, I decided to remove the center of mass motion of the 
> ion, and steer the water towards it. But then it apprears that the 
> effective steering velocity is not correct, as if it was not affected by 
> the CMM removal.

I'm not sure what you mean here. The AFM pulling works by moving a
spring at a constant rate. Currently it doesn't move with the reference
group.

> 3) Using "starting structures", it seems that the translation_rate 
> parameter is not in nm/step, because the induced translation 
> unreasonably faster.

The starting structures option is likely horribly broken. We have
generally found it easier to insert the molecule of interest and then
energy minimize (optionally using position restraints or umbrella
sampling to keep the inserted molecule fixed).

> 4) Using "starting structures", is the constraint also characterized by 
> the AFM forceconstant parameter?
> 
> 5) Using "starting structures", the steered group doesn't really go to 
> the reference group, but it passes about 3,5 A from it and continues 
> further. In which direction?? - "The truth is up there" (Mulder)
> 
>   Help me out, please, I'm getting nuts with this.
> 
> Michel

I hope to be able to improve things sometime in the future. Sorry if
things aren't working for you.

Cheers,
Justin