I created the definitions for AS4 and GL4, so hopefully I can put this
thread to rest.
To summarize some of the things that others have said:
The AS4 and GL4 residues are *only* for use with constant pH. Constant pH
works *only* with GB solvation, not explicit solvent. Since the
simulations have implicit solvent and no periodic boundary conditions you
should not be adding free counter ions (if you have ions complexed with
your system, you probably still want these). This means your system will
probably have a non-zero initial charge. This is fine. The total charge on
the system will fluctuate throughout the constant pH simulation as the
protonation state changes.
With regard to why the charge is non-integer:
It's necessary for the different protonation states to have the same
backbone charges, due to 1-4 interactions in the force field. When I
created, for example, AS4, I used the side chain charges from ASP and ASH.
In most cases, the backbone charges differed, so I had to pick a single
set of backbone charges (e.g. ASH) and apply it across all the states.
This usually meant that the total charge of the states where the backbone
charges had been changed was non-integer, so I added/subtracted the
appropriate amount of charge from the beta carbon (to put it as close
to the backbone as possible). As I recall, I only carried these
calculations out to 4 or 5 decimal places, so if you have enough
titratable residues the rounding error can add up to a few thousandths of
a charge unit. I promise you there will be much much larger sources of
error in your calculations, so this deviation should be of minimal
concern.
Cheers,
John
On Tue, 14 Dec 2004, [ISO-8859-1] Annette Höglund wrote:
> Dear Shuli and Ross,
>
> I have also encountered the same problem and I guess I wonder more about
> "why" the charge is non-integer of a system.
> How can I find what is incorrectly charged?
>
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber.scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
Received on Wed Dec 15 2004 - 03:53:00 PST
