[gmx-users] Re: Pulling a CG protein

[Thomas Schlesier]

Hi Johnny,
i replied something to the pull-code of your question. But problem was,
i didn't edited the subject line...
So here is my answer (see below) with the right subject line (so that
you hopefully find it).
Hope it helps you.
Thomas

> 
> Message: 3
> Date: Thu, 30 Jul 2009 12:08:40 -0700 (PDT)
> From: "Johnny Lam" <johntus at berkeley.edu>
> Subject: Re: [gmx-users] Pulling a CG protein
> To: gmx-users at gromacs.org
> Cc: baaden at smplinux.de
> Message-ID:
>         <56441.128.32.142.53.1248980920.squirrel at calmail.berkeley.edu>
> Content-Type: text/plain;charset=iso-8859-1
> 
> Hi XAvier, Marc, and David,
> 
> Thank you so much for the reply and encouragement ;-). Please forgive me
> as I am trying to learn how to reply to the thread that I started. With
> regards to the fun discussion, it was my original intent to compare the
> results of pulling with the MARTINI forcefield (if the pull code was
> correct) with already published works on MD (using all-atomistic modeling
> of course :-)). I just wanted to know whether the pull code that I am
> using will be valid at all. If you guys can verify that would be awesome!
> Otherwise, I'd be happy to share the results with you guys if you wish.
> Again, thanks!
> 
> --Johnny
> 
> 
> -------------------------------------------------
> Johnny Lam
> ISPE Berkeley Chapter External Vice President
> Department of Bioengineering
> College of Engineering
> University of California, Berkeley
> Tel: (408) 655- 6829
> Email: johntus at berkeley.edu
> 
> 
> ----------------------------------------------------------------------
> >
> > Message: 1
> > Date: Wed, 29 Jul 2009 16:04:54 -0700 (PDT)
> > From: "Johnny Lam" <johntus at berkeley.edu>
> > Subject: [gmx-users] Pulling a CG protein
> > To: gmx-users at gromacs.org
> > Message-ID:
> >         <55658.128.32.142.63.1248908694.squirrel at calmail.berkeley.edu>
> > Content-Type: text/plain;charset=iso-8859-1
> >
> > Dear gromacs users,
> >
> > Hi, I am trying to pull apart a relatively large protein (CG using the
> > martini force field) by pulling on two groups in opposite directions. To
> > do this, I will be using the following .mdp file. However, I am almost
> > certain that it contains errors:
> >

-snip

> >
> > ; Pulling
> > pull                =  afm
> > pull_geometry       =  direction
> > pull_start          =  no
> > pull_nstxout        =  10
> > pull_nstfout        =  10
> > pull_ngroups        =  2
> > pull_group0         =
> > pull_group1         =  pull
> > pull_vec1           =  -0.1764 -0.9823 -0.0625
> > pull_init1          =  -0.1764 -0.9823 -0.0625
> > pull_rate1          =  0.0001
> > pull_k1             =  1000
> > pull_group2         =  freeze
> > pull_vec2           =  0.1764 0.9823 0.0625
> > pull_init2          =  0.1764 0.9823 0.0625
> > pull_rate2          =  0.0000000001
> > pull_k2             =  5000
> >
> >

Think the first problem is the:
pull = afm
In the manual there is no more a 'afm' option, but as far as i know
'umbrella' should be the same (don't ask me why there was no error
message about this). If you have no pull_group0 you could get problem
with the com-motion.
If you want to pull 'pull' away from 'freeze', i would do the following:

pull = umbrella
pull_geometry = position
pull_start = yes  (we don't want to calculate the initial distance
between 'pull' and 'freeze' by hand)
pull_ngroups = 1 (because the reference group doesn't count to this value)
pull_dim = Y Y Y (because we want to pull in 3D)
pull_group0 = freeze
pull_group1 = pull
pull_vec1 = vector from 'freeze' to 'pull'
pull_init1 = 0.0 0.0 0.0

Now you are pulling 'pull' away from 'freeze' and and you should have no
problem with com-motion.
If you want to fix the position of 'freeze' i would use position_restraints.
One thing: once i had problems with the 'pull_weightsX', i got (with one
system) every time error messages (i had only 1atom in each group and
tried the values 0 and 1 both, but didn't work). So i left these to
values blank and GROMACS selected them, and it worked. In other
simulations with a similar setup i had no problems with 'pull_weightsX'.


> > The reason why group 2 has such a high force constant and low pull
rate is
> > because I wanted to simulate putting a harmonic constraint on the freeze
> > group. However, when I process this .mdp with grompp, I get the
following
> > message:
> >
> > WARNING 1 [file md_vinculin.mdp, line unknown]:
> >   Unknown or double left-hand 'pull_group2' in parameter file

Another thing could be that you have no index group with the name 'freeze'?

But best have also a look in the new (GROMACS 4) manual. COM pulling is
described from page 156 on.

Hope this helps
Thomas

> >
> >
> >
> > WARNING 2 [file md_vinculin.mdp, line unknown]:
> >   Unknown or double left-hand 'pull_vec2' in parameter file
> >
> >
> > The version of gromacs I have running on my powerpc is 4.0.5 so the pull
> > code should be implemented in the .mdp file. I am not sure if I
specified
> > the parameters correctly. Please help! Thanks!
> >
> > --Johnny
> >
> > -------------------------------------------------
> > Johnny Lam
> > ISPE Berkeley Chapter External Vice President
> > Department of Bioengineering
> > College of Engineering
> > University of California, Berkeley
> > Tel: (408) 655- 6829
> > Email: johntus at berkeley.edu
> >