[gmx-users] PMF of ligand transport

Chris Neale chris.neale at utoronto.ca
Mon May 10 18:01:59 CEST 2010


Don't use a dummy particle for your PMF if you can avoid it and test it thoroughly if you must use it. In my own work, I have received what appears to be erroneous results when using a dummy atoms together with dihedral restraints and the pull code. In my case I was actually using a dummy atom whose position was based only on other dummy atoms and I was also using angle restraints. Nevertheless, there is no reason a priori to suspect that angle restraints + a dummy based on dummies would cause problems and I wasted nearly a month figuring that out. Thomas's idea seems well-founded to me, and I'm only saying that it may not behave as you expect. Whether this is my own error or a gromacs issue or unexpected but mathematically correct behaviour is not something that I ever figured out. I'm just saying be careful.

Chris.

-- original message --

Hi Justin,

I am sorry, i dint understand this point "Your problem was only related to
the actual umbrella sampling itself." ? You meant I just want to repeat
umbrella sampling only?

In order to further understand that "Distance " problem, May I ask one more
question,( please ignore if its a stupid one..)
As Thomas suggested, = is the channel and x the dummy and l the ligand
x l =====
later time
x   ==l==
If this is the case, then  there wont be sign problem with "distance",
right???  (Once again forgive me if I am wrong, because I know you already
put your maximum effort to make me understand this thing)

 And if we use a dummy particle, will the movement  of that particle will
also matter ? How can we fix that?

Thank you.
Aswathy

On Mon, May 10, 2010 at 8:45 PM, Thomas Schlesier <schlesi at uni-mainz.de>wrote:


> > Would it be possible to use a dummy particle near the entrance of the
> > channel? = is the channel and x the dummy and l the ligand
> > x l =====
> > later time
> > x   ==l==
> > Give the dummy no charge and zero lj-parameters and construt the position
> > form the atoms at the entry of the channel.
> > I think it will be important that the dummy is farther away from the
> > channel then your ligand, else you will get problem with the distance
> > geometry (because the distance changes the sign).
> > The other important thing is that the box is long enough that the distance
> > between the dummy and your ligand is always smaller then half the box
> > vector, or else the distance goes wrong.
> >
> > (My first idea was to use the entrance area of the channel as the
> > reference, but then there would be the problems with the sign-changing of
> > the distance -> so the idea with the dummy particle).
> >
> > Greetings
> > Thomas
> >
> >
> >
> >
> >
> > Aswathy wrote:
>   
>>> > > >
>>> > > > Thank you very much for all replies.
>>> > > >
>>> > > > No w I just want to try SMd with geometry type as position . But one
>>> > > > thing is still confusing me, ie; reference group.
>>> > > >
>>>       
> >
> > If you have already generated a suitable set of positions from which you
> > have
> > generated umbrella sampling windows, you do not need to repeat the SMD.
> >  Your
> > problem was only related to the actual umbrella sampling itself.
> >
>   
>>> > > >  I have the ligand in the solvent (at the mouth of the
>>> > > > channel(extracellular).) Now I want to pull this ligand to the
>>> > > > intracellular solvent through the channel. My understanding is that the
>>> > > > reference group should be the in the same direction of the channel(ie,
>>> > > > at the intracellular end of the channel), so that if I use "position",
>>> > > > the ligand should move towards the reference group (Please correct me
>>>       
> > if
>   
>>> > > > this wrong). i have only solvent at this end(end of the channel) . Can
>>>       
> > I
>   
>>> > > > set water molecules as reference point? or any molecule at this
>>> > > > intracellular end is fine?
>>> > > >
>>>       
> >
> > In theory, you can set whatever you want, but if those water molecules
> > diffuse,
> > then you're trying to hit a rapidly-moving target!  Always choose a
> > relatively
> > static part of the structure in the direction you want to pull.
> >
> > -Justin
> >
> > --
> > 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/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/mailing_lists/users.php
> >
>   



More information about the gromacs.org_gmx-users mailing list