[gmx-users] Re: Error during CNP simulation

[Bharath K. Srikanth]

Hi Chris,

You're right, my resizing from 7.5 nm to 15 nm refers to the dimension
normal to the lipid bilayer (in this case, the x dimension)

My bilayer center of mass is at around +3 nm, and I placed the CNP at
around 10 nm to start with. I had periodic boundary conditions employed in
all three dimensions. I didn't add the CNP using genbox or any other
command- I manually manipulated the .gro coordinate file of the bilayer.
When I ran an energy minimization of just the bilayer and the CNP, the
nanoparticle moved, but very slightly- so far, so good. But, once I added
water and did another energy minimization and checked the structure after
convergence, the nanoparticle had drifted all the way up to around 0 nm,
which meant it was only around 1 nm away from my bilayer before the MD
run. So, yes, it drifted during the energy minimization.

I should probably try what you said. I repeated the entire above process
using a posre.itp file. I had created this posre.itp file using the the
lipids in the bilayer, and the CNP atoms, and I used this file during the
energy minimization after adding water, which is when i got the following
error- "A charge group moved too far between two domain decomposition
steps.".

I also got a warning before the fatal EM, which said that the number of
constraints I had imposed on the CNP (120) is greater than the number of
possible degrees of freedom (48), which could cause the system to be
unstable. I wasn't sure how to fix that, though.

I'm sorry if I'm a bit vague, but I'd appreciate if you could advice me
more on this issue.

Thanks!
Bharath

Message: 7
Date: Sun, 28 Oct 2012 19:50:11 +0000
From: Christopher Neale <chris.neale at mail.utoronto.ca>
Subject: [gmx-users] Error during CNP simulation
To: "gmx-users at gromacs.org" <gmx-users at gromacs.org>
Message-ID:
        <38D83DCC07545649B56395E217BA5E803270939C at SN2PRD0310MB382.namprd03.prod.outlook.com>

Content-Type: text/plain; charset="us-ascii"

I presume that your 7.5 nm -> 15 nm resizing refers to the Cartesian
dimension that
lies along the bilayer normal?

Also, are you sure that the nanoparticle drifts significantly closer to
the bilayer
"during minimization", as you stated? That sounds unlikely. Perhaps this
drift
occurred during MD simulation after energy minimization?

How did you implement your restraints? My first try would be using the
pull code
harmonic restraints to restrain the distance between the center of mass of
the
bilayer and the center of mass of the nanoparticle, only applying the
force along
the dimension of your global bilayer normal. Note that, when doing this, the
distance must always remain less than 1/2 of the box length along that
dimension or
your simulation will be unstable.

How did the system "blow up"? Was it a LINCS-based crash or massive
expansion of the
box or just that the bilayer fell apart? Did it blow up during energy
minimization
or MD simulation?

Chris.

-- original message --

I'm trying to simulate the diffusion of a coarse-grained carbon
nanoparticle (from Monticelli) into a coarse-grained DOPC lipid bilayer,
to reproduce the results obtained by Wong-Ekkabut et al. I first assembled
my bilayer in a box of size 7.5 x 7.5 x 7.5 nm3, and the results appeared
legitimate. However, I'm now trying to repeat the experiment by placing
the carbon nanoparticle at an initial position further away from the
bilayer, and so I've resized my box to 15 nm x 7.5 nm x 7.5 nm.

In this case, what happens is that after solvation using genbox, if I
carry out an energy minimization, the CNP drifts very close to the
bilayer, rendering my increase of the box size redundant. To overcome
this, I've tried restraining the positions of the CNP and the bilayer, but
when I do this, my simulation 'blows up'. I'm not aware of how to overcome
this problem. Would you be able to help me with this?

Thanks!

Regards
Bharath K Srikanth