Re: Membrane Protein Simulation Problems

From: Marcos Sotomayor (
Date: Sat Oct 13 2007 - 11:17:49 CDT

Just to clear up the confusion (I might be wrong though)

-The protein forcfield is the CHARM22 forcefield. It has
not changed at all in the version 27 or 31 of CHARMM, except that in CHARMM31
they provided the CMAP corrections, which you should apply on top of the
CHARM22 force-field.

-The difference between CHARM22 and CHARM27 is in the lipid
parameters, which are not affected by CMAP.

-So, just to be sure that you are using the parameter and topology
files provided in the version 31 (or >31) of CHARMM (the software) and
that your psf file does include the CMAP terms and that NAMD is actually
using the CMAP correction: check CROSSTERMS in the log file (you should
have more than 6 under STRUCTURE SUMMARY).

I hope I did not confuse people even more....

Two more comments for Ilya: (1) if you are applying patches during psf
file generation (outside the generate statement), be sure to use the
regenerate angle dihedrals command [This is important when using
topology files from c31b1 and c32b1] (2) be sure to align your protein
before computing RMSDs.


On Sat, 13 Oct 2007, L. Michel Espinoza-Fonseca wrote:

>>> My understanding from the previous email is that 27 has the CMAP corrections
>>> as well.
>> 22 and 27 do not include CMAP, which was released with 31. However, the
>> labeling is quite confusing and I see that you seem to be using the right
>> parameter file (which is labeled 27 and includes CMAP!?). Check the number
>> of CROSSTERMS in your log file (STRUCTURE SUMMARY section) to be sure
>> that you are using CMAP.
> Just a single comment:
> CHARMM 27 *does* include the CMAP correction. When using this version
> of the ff, your NAMD log file shows that you have "x" number of
> crossterms. I've been using the CHARMM 27 ff for the stability of
> helices and you can indeed see the difference in terms of stability.
> In fact, the paper by Buck et al. mentions such correction as
> "C22/CMAP".
>>> "reinitvels" must not be working because I do not assign a temperature, thus
>>> it would not know what distribution to select from. I have since removed
>>> this option. Not even sure how it got into my script.
>> Indeed, you are not using the "temperature" command, but your posted
>> configuration files has a "set temperature 310" command and
>> a "reinitvels" command (the command does work and you were overriding
>> velocities in that particular case; note the subtle difference between
>> the TCL variable temperature and the NAMD command temperature).
>>> Temperature averages out to 310K. The PME grid is 128x128x128.
>> What is the size of the simulation box?? Is there enough space between
>> periodic images?
>>> 1.4 A resolution.
>> Is the whole protein or a particular zone getting deformed? you may
>> want to compute RMSD for some zones of the protein or compute RMSD per
>> residue and check what part of the protein is behaving badly. By the
>> way, when you said that your RMSD was going up, how bad is it? 2, 3, 6 A?
>> I would also suggest to look at protonation states carefully (as suggested
>> by Richard).
>>> Hope that helps,
>>>> Marcos

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