Re: Membrane Protein Simulation Problems

From: Marcos Sotomayor (sotomayo_at_ks.uiuc.edu)
Date: Sat Oct 13 2007 - 10:07:59 CDT

Hi Ilya,

Looking at your previous e-mails, I would suggest the following:

- Use the CHARMM31 force-field, which is the same as 22 and 27 for
proteins but also includes the CMAP correction. The CMAP correction has
been shown to stabilize RMSD and alpha-helices (see Buck et al, BiophysJ
Letter, 90:L36-L38 2006). You will need to recreate your psf files.

- I wouldn't use the "zeromomentum" option when using constraints and/or
Langevin dynamics, as in those cases the system *should not* conserve
momentum (it is OK if the center of mass of the protein moves around a
bit).

- Do not use the "reinitvels" command every time you restart a simulation,
as you are reading the velocities from the previous run (in the conf file
you posted before, you were basically reading the velocities from a
previous run and then overriding them).

- Is your system neutral overall?

- Check your periodic boundary conditions, your temperature, the size
of your PME grid.

- If your protein is a bad model, or of poor resolution, or somehow
distorted by crystallographic artifacts, you may never get a good
convergence on RMSD. Unfortunately MD will not refold your protein on a 10
nanosecond time scale :-)

Hope that helps,
Marcos

On Fri, 12 Oct 2007, Ilya Chorny wrote:

> Dear Richard,
>
> I got your contact info from the NAMD mailing list. I am a member of the
> Stroud Lab at UC San Francisco and I am working on a membrane protein
> simulation. I am having some trouble with my protein. I prepare my
> simulation box by inserting the protein into the membrane, deleting
> overlapping lipids, solvating the system and adding counter ions. I minimize
> and then equilibrate the system for 2 ns. In the initial 2 ns simulation I
> fix the position of protein atoms and constrain the lipid phosphates in the
> Z dimension (along the pore axis). I then run an 8ns simulation in which I
> release the constraints on the lipid and constrain the CA positions of the
> protein. After my 10 ns equilibration I run another 8 ns simulation in which
> the whole system is free. In the last 8 ns simulation my protein seems to be
> falling apart. The RMSD keeps increasing and the channel residues which
> should be fairly constrained move all over the place. I monitor my
> equilibration by calculating the protein-lipid energy which seems to have
> converged. Any ideas on what I can do to debug my problem? The membrane is
> POPE.
>
>
> Thanks,
>
> Ilya
>

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:45:22 CST