Re: [AMBER] large resp charges for non-natural amino acid

From: Arturas Ziemys <arturas.ziemys.uth.tmc.edu>
Date: Thu, 29 Jan 2009 09:10:50 -0600

Hi, Francois,

First, thanks a lot for your replay.

My system is just a simple peptide that binds into another protein. So,
I need just one independent modified amino acid to develop. Initial
conformation I have chosen to minimize with Gaussian was manually
created to cover essential geometry changes could happen. To be precise,
the number of ESP points is much less than in A1 (new) tutorial - I used
Gaussian run based on antechamber creates Gaussian input. The number of
ESP points is ~8000 and 22 atoms. This alpha-carboxyl-glutamic acid has
charge -2. I guess this is the source of all level, but maybe I'm wrong,
because I try first time to go for charge derivation for amber FF. I
want to stick to standard ff99SB as close as possible in charge scheme.
One important point - I have used resp within AMBER 9 distro.

After many many different tries playing with resp modifying fitting
procedure, I could not get any reasonable charge on CA. It was the only
atom which was spoiling all fitting. Also tried to target fitting
charges to Mulliken from Gaussian output - helped not much. As far as
amber FF tends to use fitted charges, it shoul be used all residue (not
like e.g. CHARMM where you can have sero-charge fragments). Something
worked than I compared my residue to standard GLU and ASP. Along with
caps, I have frozen NH, CO to GLU and ASP values (they are the same).
ALSO, I manually put charge -0.05 on CA (something more negative than
standard GLU/ASP as my reside is -2 e). After playing with restrains I
managed to get "rational" charge scheme. Using this approach my
aliphatics hydrogens (CB,...) are positive and carbons negative
(physically expected, but in opposite to ff99). Maybe amber community
will not crucify for this approach :)

Gaussian outputs are attached in tar.gz.

best
Arturas

P.S.-1I have checked AMBER mailing list regarding resp - tons of
messages. I could not find something direct what is the *standard*
procedure to develop charges for amino acids ?

P.S.-2 sorry for possibly repeated message, but amber list was not
accepting few tries to send replies.


FyD wrote:
> Dear Arturas,
>
>> I have performed 3 conformation RESP fitting to calculated charges for
>> alpha-carboxyl-glutamic acid. I have used as a basis AMBER advance
>> tutorial about fluorescein and linker (tutorial A1). What I did I
>> capped my residue with ACE and NME groups. Minimized 3 conformations.
>> Restrained cap charges as said in the tutorial.
>>
>> Everything works fine, except CA gets charge 1.2 after RESP fitting.
>> Also other atoms in peptidic fragment have larger charges.
>
> It is difficult to answer as you do not provide a lot of information
> about the molecular system you are working on. If you get a charge of
> 1.2 for a CA atom something wrong appends in your strategy, I guess.
>
> There were some intense discussions about this tutorial & I do not
> want to reignite any tension. However, you have to be aware that the
> fitting step in this tutorial is not correct. This tutorial should
> better define the parameters used to help new users to understand the
> choices made. This tutorial will be corrected.
>
>> To increase RESP restrain factor ?
>
> You can try, but I do not think so.
>
>> Increase number of points in ESP calculations?
>
> The number of points is already very high in this tutorial, and in
> such a case it is not really useful (although you never know as we do
> not know very well the type of structures you are working on).
>
>> Or maybe for amino acids (ff99SB) there is another protocol to
>> develop charges ?
>
> There is one charge derivation procedure for AMBER force field (that
> can be modified when it does not work in some specific cases). My
> feeling is that in your case the default charge derivation procedure
> for AMBER force field should work. I might be wrong but I think you
> made some errors and/or the fitting step of this tutorial generate
> artefacts.
>
>> Any suggestion or comments ?
>
> If you want I can run R.E.D. Server for you, apply the AMBER default
> charge derivation procedure and see if it works for you. In this case,
> just send to my personal email these 3 geometry optimization output
> files and I will generate a 4-molecules charge fit using either each
> of your conformations taken individually and your three conformations
> taken as a single molecule. This will take 10 minutes.
>
> I hope this helps,
> regards, Francois
>
>
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>


-- 
Arturas Ziemys, PhD
  School of Health Information Sciences
  University of Texas Health Science Center at Houston
  7000 Fannin, Suit 880
  Houston, TX 77030
  Phone: (713) 500-3975
  Fax:   (713) 500-3929  



_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber

Received on Fri Jan 30 2009 - 01:36:59 PST
Custom Search