Re: [AMBER] paramfit questions

From: Karl N. Kirschner <kkirsch.scai.fraunhofer.de>
Date: Fri, 19 Apr 2013 09:26:09 +0200 (CEST)

Hello Cody,

  Breaking the molecule up into smaller analogues is a good idea. R.E.D. and its website are capable of generating charges for these analogues as if they were residues for the parent molecule. I believe there are some examples/tutorial for doing this at the R.E.D. website. One issue that you might have for trying to determine the charges for such a large molecule on a single go, is that the input conformation will have a noticeable influence on the resulting partial atomic charges. It is easier to scan for low energy conformation of smaller molecules that together can be used to determine an average charge set. The additional advantage of R.E.D. is that resulting partial atomic charges become more reproducible by others, and if you submit it to their molecular database then others can take advantage of the results.

  For the internal coordinate parameterization, you always want to compare "apples-to-apples," meaning that if you create a PE curve using QM calculations that constrain a specific dihedral, then you want to do the same thing in the MM calculation. Concerning the MM calculations, you want to set the force constants of the dihedral angle in question to zero and obtain the resulting PE curve - this gives you a baseline for which Vn terms to add to bring the MM curve into agreement with the QM curve.

  You might also want to look at the Wolf2Pack website. There is some information concerning force field parameterization topics that you might find helpful if you are new to the topic. This site will also help you optimize bond, angle, and dihedral force constants that are compatible with Parm99SB, gaff, and Glycam06. W2P will not automatically provide you with optimized parameters, like parmfit will do, but you will have to adjust them by hand (which is possibly a good thing to do). Quite a bit of already computed QM data for internal coordinates are available for using as a target observable; when an appropriate molecule or QM are not present, just request it.

Best regards,
Karl

RED: http://q4md-forcefieldtools.org/RED
W2P: www.wolf2pack.com

----- Original Message -----
From: "Cody Lance Covington" <cody.l.covington.Vanderbilt.Edu>
To: amber.ambermd.org
Sent: Thursday, April 18, 2013 7:42:24 PM
Subject: [AMBER] paramfit questions

Hi

I was looking into the paramfit program to get a few dihedral parameters, and had a few questions that I haven't found answers to in the manual or mailing list.


1) The molecule I am parameterizing has 60 atoms so I was going to try a smaller fragment. Since electrostatic interactions are calculated in 1-4 interactions, are there any considerations for RESP charges? I could use the entire molecule and lower the basis set.

2) The Amber force field will have a different set of equilibrium bond lengths and angles than the QM calculations. Is this accounted for in the 'K' term? Shouldn't a geometry optimization be performed in amber with the ff with the dihedral angle fixed, then a QM optimization calculation with the dihedral angle fixed? From the examples in the manual, it looks like the energies are just calculated using both MM and QM and compared.
Any help would be appreciated!

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

_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Fri Apr 19 2013 - 00:30:02 PDT
Custom Search