AMBER Archive (2009)

Subject: Re: [AMBER] R.E.D intra-mcc and reorientation

From: FyD (fyd_at_q4md-forcefieldtools.org)
Date: Sat Jul 11 2009 - 03:05:56 CDT


Dear Ganesh,

> I am interested in partial charge derivation of a polymer. I am
> using R.E.D III for this. I have a few questions regarding the set up:
>
> 1) I decided to derive the partial charge of one monomer and use
> those charges for each constituent monomer in the polymer. I was
> wondering if people could comment on the advantages and
> disadvantages of this procedure.

You are right this is a key question: The basis of the answer is the
work of Kollman's group published in 1995 by Cieplak et al.
Application of the multimolecule and multiconformational RESP
methodology to biopolymers: Charge derivation for DNA, RNA, and
proteins. J. Comput. Chem. 1995, 16, 1357-1377.

The basic idea behind this work is instead of computing the charges
for a (big) polymer such a polypeptide, polysaccharide or
oligonucleotide (that cannot be handled by quantum mechanics because
too big, anyway), to derive charges for the repeating units
constituting the polymer.

The advantages:
- You are indeed able to derive the charges for macrosystems based on
building block units & force field topology database (FFTopDB)
building: this is the basis of _ALL_ empirical force fields nowadays.
- You can potentially build any type of macrostructures using this
approach; See for instance:
http://q4md-forcefieldtools.org/Tutorial/Tutorial-3.php
http://q4md-forcefieldtools.org/Tutorial/Tutorial-3.php#22
http://q4md-forcefieldtools.org/Tutorial/Tutorial-3.php#27
   and
http://q4md-forcefieldtools.org/Tutorial/Tutorial-3.php#26 for an
organo-glycoconjugate.
- You rigorously control the conformation(s) of each unit used in
charge derivation (conformations does affect charge values based on
MEP computation).
- When deriving charge for large structures you often select the
conformation based on luck/random_?. I do not like the idea of luck
deciding for me ;-)

The disadvantages:
- The procedure appears complex for new users. (R.E.D. has been
develop for this purpose to make the work of Cieplak et al. more
accessible to new users. We have written several tutorials to help as
well. See http://q4md-forcefieldtools.org/Tutorial/)
- When generating the building block units, constraints during the
charge fitting step for "connecting group(s)" are introduced. Here, if
you do not know which chemical group you want to use as a connecting
group, you will introduce errors in the fitting step (that are
difficult to judge/quantify). R.E.D. IV incorporate now a statistics
module to solve this problem. This new feature will be first released
through R.E.D. Server (soon).

> 2) Since a monomer is a fragment and not a complete molecule, I
> "artificially" satisfied the valence by bonding it to other groups.
> I then set an intra-mcc to derive charges only for the relevant
> fragment. As I understand, this is in accordance with the R.E.D III
> tutorial
> (http://q4md-forcefieldtools.org/Tutorial/Tutorial-1.php#10). I was
> curious as to how the choice of the group selected to satisfy the
> valence will affect the partial charge.
> eg. Let us say the polymer is (-CH2-)n. I derive partial charges for
> CH3-CH2-CH3 with intra-mcc on both CH3s. Would this provide
> accurate and useful partial charges? What factors would govern the
> choice of the chosen group (CH3 in this case)?

You are 100 % right... Once again the publication by Cieplak et al.
should provide you a starting point for your polymer.
Our http://q4md-forcefieldtools.org/Tutorial/ tutorials should help
also. R.E.DD.B. has been totally rebuilt; we will submit new projects
that will present examples different from proteins & nucleic acids.
Finally, we can provide you private assistance for your project as
well, if you use the "Help" service available from R.E.D. Server.

Your CH3-CH2-CH3 case is "ridiculously" small, but perfectly well
chosen. "ridiculously" because too small, but is a perfect example
because you need to exclude the two CH3 chemical groups to construct a
polymer of CH2s i. e. (CH2)n. In this case, using INTRA-MCCs set to
zero for these 2 methyl groups should work and introduce small errors
during the fitting step...

> 3) Regarding RBRA reorientation
> (http://q4md-forcefieldtools.org/Tutorial/Tutorial-1.php#3), I was
> wondering if the choice of the 3 atoms is arbitrary. In other words,
> can I choose any 3 atoms within my molecule for the reorientation
> procedure?

You are right once again: Nothing else to say. Now you could select:
- only heavy atoms,
- heavy atoms common to all the units constituting your polymer,
- more generally if you reorient once using the atoms "1 2 3", we
always reorient the same structure a second time using "3 2 1": This
allows cancelling out the (small) effect of the 1st orientation by the
2nd orientation. This was clearly quantified for the dimethylphophate
molecule for instance when building the nucleic acid FFTopDB.

Thanks for this discussion. May be Piotr would like to add something...

regards, Francois

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