[gmx-users] Fwd: Last step before CG em.mdp

[Justin A. Lemkul]

Francesco Pietra wrote:
> On Mon, Nov 30, 2009 at 8:18 PM, Justin A. Lemkul <jalemkul at vt.edu> wrote:
>>
>> Francesco Pietra wrote:
>>> Failed to specify that both the DPPC bilayer and the box of water are
>>> from equilibrated systems in MARTIN web, while the protein was Amber
>>> minimized (and MD) before generating the CG.
>>>
>>> I have tried the cg protein alone
>>>
>>> genbox -cp my.pdb -box 15 -o my.gro
>>>
>> Again I ask why you are using SPC as the solvent.  If you indeed want to use
>> the MARTINI CG water representation, you must pass its configuration
>> explicitly to the -cs flag.  Otherwise, the default (spc216.gro) is used.
> 
> The water that I set around the extracellular part of the protein was
> equilibrated cg water (by just multiplying the box furnished on
> MARTINI). In relation to the command above, I don't understand your
> remark, simply because there is no water at all, both in the input
> .pdb and output.gro. This is true for my previous post, too.
> 

Oops, my mistake.  I forgot that -cs was optional, and not taken as default :) 
Never mind that comment.

> 
>> Are you defining the protein's position in the box before solvation?  Using
>> editconf -c is your friend.
> 
> I did present work before getting your mail. Curiously, gmail now
> places your mail in spam. Luckily I noticed that.
> A point I have to pay much attention to - for a possible major mistake
> in my procedure - is your "before solvation". I assumed no solvation
> at all.
> 
> 
> 
>>> grompp -f em.mdp -c my.gro -p my.top -o topol.tpr
>>>
>>> mdrun -v -s topol.tpr
>>>
>>> steepest descents converged to machine precision but not to the
>>> requested Fmax < 2000
>>>
>>> Pot Ener 7.1e+08
>>> Max force 6.1e+10 on atom 983
>>> Norm of force 2.5e+11
>>>
>>> confout.gro shows that one of the subunits has been displaced from the
>>> multimer (apparently both intact) There is a long bond between the
>>> displaced subunit and the rest of the multimer, very closely to the
>>> atom 983 of max force, located in the pore region. I have scarce
>>> experience with gromacs yet, hope that this suggests remedies. I used
>>> a crude version of em.mdp.
>>>
>> Well, there are several potential problems.  One is the box size (as I
>> mentioned in my last message).  The other question is whether or not the
>> conversion from atomistic to CG was done correctly.  I know I have mentioned
>> the problems with the MARTINI script a number of times, and perhaps this has
>> also come up in our discussions before, but do check that the starting CG
>> coordinates make sense.
> 
> The cg protein model superimposes nicely to the full-atom model. The
> dimensions I gave in my previous mail were DPPC and W inclusive but
> "-box 15" is still to small. The protein might have seen its neighbor.
> 

Then that's certainly the first problem to fix.  This is why I generally use 
editconf for defining the box (and why I keep recommending it).  For a simple 
protein in water, using a command like:

editconf -c -d 1.0

will define a suitably-sized box so that you don't have to guess at the box 
dimensions.  There is no such option with genbox.

>> Beyond that, try an in vacuo minimization of your protein and see what
>> happens there.
> 
> Do you mean without setting "-box"? I got confused about that, I
> thought to have vacum around the protein (or the protein+DPPC+W of
> previous mail) with
> 

You can disregard that comment, I had assumed from reading the earlier post that 
you had solvated the structure.  If the protein does not minimize by itself 
(hence, in vacuo) then perhaps your issue comes from your box dimensions.  Seems 
most likely at this point.

I don't think you want a vacuum around your system.  That's the whole point of 
PBC is it not - to avoid boundary conditions and "droplet" formation?  What you 
want to do is assign a box that actually fits your system :)

-Justin

-- 
========================================

Justin A. Lemkul
Ph.D. Candidate
ICTAS Doctoral Scholar
MILES-IGERT Trainee
Department of Biochemistry
Virginia Tech
Blacksburg, VA
jalemkul[at]vt.edu | (540) 231-9080
http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin

========================================