AMBER Archive (2009)

Subject: Re: [AMBER] Possibility to use ff99 + GAFF + GLYCAM_06 in one time for complex system ? - Last detail

From: Karl Kirschner (kkirsch_at_scai.fraunhofer.de)
Date: Thu Apr 23 2009 - 08:54:14 CDT

Hi Marek,

        Your solution is as good as any. I must reiterate my caution of creating a
new force field via the Frankenstein method. Each force field that is
developed has been done so in a specific and unique manner. Since there are
several possible parameter solution for any given force field term, and the
fact that these terms are usually dependent (coupled) upon other terms in the
force field, combining them in this manner can lead to
unpredictable/unreliable behavior.

        I have used this method myself, but on fairly rigid systems where I don't
expect much flexibility to occur during the dynamics. I have also validated
the resulting parameters the best I could by comparison to some experimental
known observable.

Cheers,
Karl

On Thursday 23 April 2009 15:41, Marek Malư wrote:
> Dear Karl,
>
> thanks a lot again !
>
> There is obviously one last question:
> Where to take the numbers for a "new" gaff
> terms n3-CG, n3-CG-CG , n3-CG-CG-CG ...
>
> One of the natural solutions is in my opinion this:
>
> n3-CG = n3-c3
> n3-CG-CG = n3-c3-c3
> n3-CG-CG-CG = n3-c3-c3-c3
>
> etc.
>
> If I am wrong, please correct me, otherwise do not waste time.
>
> Thanks for all !
>
> Best,
>
> Marek
>
>
>
>
>
> Dne Thu, 23 Apr 2009 13:43:31 +0200 Karl Kirschner
>
> <kkirsch_at_scai.fraunhofer.de> napsal/-a:
> > Hi Marek,
> > Attempted answers below :) .
> >
> > On Wednesday 22 April 2009 04:31, Marek Malư wrote:
> >> Dear Karl,
> >>
> >> thanks again !
> >>
> >> It seems to me a little more complicated than on the start :)) but I am
> >> not going to gave up it yet.
> >> In any case I can always use the "cheaper" alternative with using GAFF
> >> for dendrimer and also for my bonded maltose, although the maltose
> >> topology
> >> will not be probably optimal in this case ...
> >>
> >> OK so back to our little complicated story:
> >>
> >>
> >> #1
> >> Conversion of GAFF atom types to GLYCAM ones in generated prepin files
> >> couldn't be problem however also here could be some small unclearnesses
> >> for example to convert c3 I have two possibilities c3->CG or c3->CT .
> >> CG and CT has the same description ( "sp3 C aliphatic" ) in
> >> Glycam_06c.dat
> >> so
> >> what is the difference between them ?
> >
> > As you know c3, CT, and CG are all atom types that refer to sp3 C
> > aliphatic.
> > These differences occur because each parameter set was derived in a
> > unique
> > way. Giving them different atom names help to maintain that parameters
> > from
> > each force field do not overwrite the other's parameters. For Glycam, by
> > changing only the sp3 C aliphatic atom type, we could ensure that all of
> > Glycam parameters do not overwrite any protein or rna/dna parameters if a
> > user wishes to model a mixed bio system.
> >
> >> #2
> >> The most problematic point I see here on the "ff discontinuity" which
> >> you
> >> mentioned in your last email. I am not sure if I understood perfectly
> >> what is necessary to do to solve this problem. So I illustrated
> >> my understanding in attached picture. Please let me know if I am
> >> following your ideas in the right manner or not.
> >
> > Your illustration represents this idea for the particular bond in
> > question,
> > but there are also a few angle bending (~4 terms) and several torsion
> > terms
> > (maybe 7 or so) that will need attention.
> >
> >> If yes please let me know if I forgot for some parameters which
> >> should be also add/change to/in gaff.dat and also maybe in some other
> >> amber files (problem of dependencies) also regarding to later MM-PBSA
> >> analysis etc.
> >
> > I think you are forgeting the angle and torsion terms involving hydrogen
> > atoms, plus n3-CG-CG-OH
> >
> >> #3
> >>
> >> What is the difference between files glycam04.dat, glycam04EP,
> >> Glycam_06.dat, Glycam_06a.dat,
> >> Glycam_06b.dat , Glycam_06c.dat , Glycam_06d.dat , Glycam_06e.dat ,
> >> Glycam_06f.dat ?
> >
> > The difference between glycam04.dat, glycam04EP is the existence of Extra
> > Points that resemble lone electron pairs. The differences between the
> > rest
> > are just new versions of the force field where some parameters have been
> > changed to hopefully correct an error in the force field. Glycam_06.dat
> > and
> > Glycam_06a.dat are the originals and Glycam_06f.dat is the most recent in
> > your list. You can access the most recent Glycam force field files at
> > http://glycam.ccrc.uga.edu
> >
> > Cheers,
> > Karl
> >
> >> Thanks in advance for your comments or of course for comments of anybody
> >> else !
> >>
> >> Best,
> >>
> >> Marek
> >>
> >>
> >>
> >>
> >>
> >>
> >>
> >>
> >>
> >> Dne Tue, 21 Apr 2009 09:24:18 +0200 Karl Kirschner
> >>
> >> <kkirsch_at_scai.fraunhofer.de> napsal/-a:
> >> > Hi Marek,
> >> >
> >> > I believe that Glycam_06 possesses the necessary parameters for your
> >> > broken
> >> > glucose ring residue (maltose already has a prep file). However, and I
> >> > forgot
> >> > to mention this, you will have to develop a prep file for this new
> >> > residue
> >> > and the appropriate charges that are used with Glycam_06. You can use
> >> > antechamber to create a prep files for the broken glucose ring
> >>
> >> residue,
> >>
> >> > using
> >> > the gaff force field, and then modify the atom types by hand to be
> >> > Glycam_06
> >> > atom types. The charges can be computed using the Resp algorithm in a
> >> > traditional sense as published in many papers (see J. Phys. Chem. 97,
> >> > 10269
> >> > (1993); J. Am. Chem. Soc. 117, 5179-5197 (1995)) or using the newer
> >> > R.E.D.
> >> > technique (http://q4md-forcefieldtools.org/RED/).
> >> >
> >> > In general, note that when you use Gaff with one of the
> >> > protein/rna/glycam
> >> > force fields/prep files, and there are covalent bonds between the
> >> > segments
> >> > that you want to model using the different force fields, you will
> >>
> >> have to
> >>
> >> > modify by hand the Gaff parameters so that the terms contain uppercase
> >> > letters to correspond to the prep files in amber's residue database
> >> > ($AMBERHOME/dat/leap/prep). Otherwise, in the areas where these
> >>
> >> residues
> >>
> >> > link
> >> > together you will have problems getting leap to assign the parameters.
> >> >
> >> > I am not sure that parmchk can currently check for correct glycam_06
> >> > parameters. You will have to do this by checking the files visually if
> >> > you
> >> > want to use glycam_06.
> >> >
> >> > Cheers,
> >> > Karl
> >> >
> >> > On Tuesday 21 April 2009 05:52, Marek Malư wrote:
> >> >> Dear Karl,
> >> >>
> >> >> thanks again ! If I understood well it is possible to use GLYCAM_06
> >>
> >> ff
> >>
> >> >> for parametrisation of my "arbitrary" carbohydrate residuum so easy
> >> >> like to use GAFF forcefield for this purpose. And moreover GLYCAM_06
> >> >> contains
> >> >> already big library of common carbohydrates in "Glycam_06.prep".
> >> >>
> >> >> If I am not wrong in this general assumption I have one more question
> >> >> for
> >> >> you ( hopefully really last one in this story :)) )
> >> >>
> >> >>
> >> >> OK, if I need to parametrise my dendrimer residui (i.e. "central",
> >> >> "repetitive" and "end" building units)
> >> >> I just use ANTECHAMBER for creation of the relevant prepin files
> >>
> >> which
> >>
> >> >> contains GAFF atom types.
> >> >>
> >> >> If the analogous way could be used for parametrisation of
> >>
> >> carbohydrate
> >>
> >> >> residuii using GLYCAM_06 ff
> >> >> how to force ANTECHAMBER to ("disable" GAFF ff) and generate PREPIN
> >>
> >> file
> >>
> >> >> of my carbohydrate residuum
> >> >> (for example my maltose with one broken glucose ring) with GLYCAM_06
> >> >> atom types (not GAFF ones ) ? and also to force consecutively PARMCHK
> >> >> routine to check/associate
> >> >> GLYCAM_06 and not GAFF forcefield parameters with respect to given
> >> >> residuum ?
> >> >>
> >> >> Thank you in advance for explanation of this very last technical
> >>
> >> detail
> >>
> >> >> !
> >> >>
> >> >> Best,
> >> >>
> >> >> Marek
> >> >>
> >> >>
> >> >>
> >> >>
> >> >> Dne Mon, 20 Apr 2009 14:27:45 +0200 Karl Kirschner
> >> >>
> >> >> <kkirsch_at_scai.fraunhofer.de> napsal/-a:
> >> >> > Hello Marek,
> >> >> >
> >> >> > You should be able to model the "broken" glucose fine using
> >> >>
> >> >> Glycam_06.
> >> >>
> >> >> > Glycam_06 was parameterized using hydrocarbons, alcohols, and
> >>
> >> ethers
> >>
> >> >> > (plus a
> >> >> > more limited number of nitrogen and RCO2- compounds; J. Comput Chem
> >> >>
> >> >> 29,
> >> >>
> >> >> > 2008,
> >> >> > 622). It was tested on a small molecule test suite and on
> >> >>
> >> >> carbohydrates.
> >> >>
> >> >> > Thus, Glycam_06 should work well for simulating the broken glucose.
> >> >> >
> >> >> > You may be missing parameters for where the glucose is linked to
> >>
> >> the
> >>
> >> >> > dendrimer. You could use parameters from either ff99 or gaff for
> >>
> >> the
> >>
> >> >> > missing
> >> >> > parameters. Due to the flexibility and the number of highly
> >>
> >> rotatable
> >>
> >> >> > bonds
> >> >> > of your system, it would be advantageous to test the resulting
> >>
> >> force
> >>
> >> >> > field
> >> >> > against experiment or QM to validate your parameter choices.
> >> >> >
> >> >> > Cheers,
> >> >> > Karl
> >> >> >
> >> >> > On Monday 20 April 2009 13:52, Marek Malư wrote:
> >> >> >> Dear all, I have very last question to this topic.
> >> >> >>
> >> >> >> As I reported sooner I would like to simulate dendrimers
> >> >> >> "decorated" with maltose.
> >> >> >>
> >> >> >> Unfortunately binding of maltose to terminal dendrimer amines
> >> >> >> requires opening of one glucose ring and creating bond between
> >> >> >> C1' and dendrimer terminal amine ( please see the attached
> >>
> >> picture ).
> >>
> >> >> >> In this case I assume that GLYCAM_06 forcefield could be
> >> >> >> used only for parametrisation of the second ( unbroken ) glucose
> >> >> >> ring and the broken one should be parametrised using GAFF ff.
> >> >> >>
> >> >> >> Am I right ?
> >> >> >>
> >> >> >>
> >> >> >> Thanks in advance for answering of this my very last
> >> >> >> question.
> >> >> >>
> >> >> >> Marek
> >> >> >
> >> >> > _______________________________________________
> >> >> > AMBER mailing list
> >> >> > AMBER_at_ambermd.org
> >> >> > http://lists.ambermd.org/mailman/listinfo/amber
> >> >> >
> >> >> > __________ Informace od NOD32 4021 (20090420) __________
> >> >> >
> >> >> > Tato zprava byla proverena antivirovym systemem NOD32.
> >> >> > http://www.nod32.cz 

-- 
------------------------------------ 
Karl N. Kirschner, Ph.D.
Fraunhofer-Institute for Algorithms
    and Scientific Computing - SCAI
Department of Simulation Engineering
Schloss Birlinghoven
53754 Sankt Augustin, Germany
Tel: +49 (0) 2241-14-2052
Fax: +49 (0) 2241-14-1328
------------------------------------ 

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