Yun,
As Francois noted, we use the MD ensemble to weight the charges towards
solution conformations. My only concern is your QM levels are a bit off.
For neutral sugars, we optimize and develop the ESP with a 6-31G* basis set
whereas for anionic systems 6-31++G** is used for optimization and ESP
calculations. The RESP weight is the same (0.01) either way and the rest of
the options you mentioned are correct. I would use a crystal geometry for
your pre-ensemble averaged charges (if available) or an optimized QM
structure.
>> " This really seems to a daunting task even with only one small
molecule."
It can be, however the payoff is better fits with experiment observables.
Typically, the MD setup and simulation takes about 24-48 hours while the
charge development can take anywhere from 24-70 hours depending on computing
resources. One of the biggest obstacles to going from the MD to QM
calculations is freezing your exocyclic rotamers. This can be done with the
opt=modredundant command in Gaussian and then listing the frozen torsions in
the torsion scan list where you do a zero-step torsion rotation.
>> " So if what is calculated from the newly-built force field is quite
different
>> from experiment, we generally still keep the charges and only modify the
>> angle and dihedral parameters with some kind of chemical intuition?"
This would be true if you knew what the solution rotameric populations were
for every exocyclic rotamer (hydroxyls, C5-C6, anomeric, etc). However
complete rotamer population information is usually beyond the scope of
current experimental methods and experimental methods usually don't tell you
how the rotamer populations are coupled together.
-Matt
-----Original Message-----
From: Yun Shi [mailto:yunshi09.gmail.com]
Sent: Monday, September 19, 2011 6:27 PM
To: AMBER Mailing List
Subject: Re: [AMBER] about Glycam_06g.dat and thiol-glycosidic linkage
Hi Matthew,
I saw "The initial partial charges for the solute were computed from the
crystal geometry" from the paper you referred. So I guess in order to obtain
initial charges, I should start with either a crystal structure, if
possible, or a QM optimized structure (at the HF 6-31G ** level?). Then use
the HF 6-31G* and CHELPG, with a constraint of 0.01, to derive initial
charges, which can be done automatically on R.E.D. server. And then do MD as
specified in glycam06 paper, and then derive ensemble-averaged RESP charges
... This really seems to a daunting task even with only one small molecule.
So if what is calculated from the newly-built force field is quite different
from experiment, we generally still keep the charges and only modify the
angle and dihedral parameters with some kind of chemical intuition?
Thanks for your explanation.
Yun
On Mon, Sep 19, 2011 at 7:41 AM, Matthew Tessier
<matthew.tessier.gmail.com>wrote:
> Yun,
>
> >> But which MM force field should I use for this MD simulation?
>
> I'm not sure what you're asking here because you're using the GLYCAM force
> field to generate the ensemble averaged charges. The initial charges you
> should use will come from a QM optimized model of a methyl-reducing
> monosaccharide with the thiol-linkage. See:
> M. Basma, S. Sundara, D. Calgan, T. Varnali, R.J. Woods. 2001. Solvated
> ensemble averaging in the calculation of partial atomic charges. J.
Comput.
> Chem. 22: 1125-1137.
>
> >> With a lack of experimental data, does that make sense to compare the
> >> minimized structure of the test molecule in solvent with the force
field
> >> parameters we are developing, with the QM optimized structure in
vacuum?
>
> I think you're asking if comparing the QM properties to the solution MD is
> a
> good comparison. The answer is no. GLYCAM is parameterized to reproduce
> solution properties which is one reason we use ensemble averaged charges.
> Additionally, the glycosidic rotation profiles can be different in
> solution.
> If no experimental data is available for your system then you should find
a
> similar thiol-linked model that has experimental data (J-couplings, NOEs,
> etc) to validate against.
>
> -----Original Message-----
> From: Yun Shi [mailto:yunshi09.gmail.com]
> Sent: Friday, September 16, 2011 12:46 PM
> To: AMBER Mailing List
> Subject: Re: [AMBER] about Glycam_06g.dat and thiol-glycosidic linkage
>
> Hi Matthew,
>
> I understand that a MD run is required to generate conformational
ensembles
> so that we could calculate ensemble-averaged RESP charges. But which MM
> force field should I use for this MD simulation?
>
> With a lack of experimental data, does that make sense to compare the
> minimized structure of the test molecule in solvent with the force field
> parameters we are developing, with the QM optimized structure in vacuum?
>
> Thanks,
> Yun
>
>
> On Thu, Sep 15, 2011 at 11:53 AM, Matthew Tessier <
> matthew.tessier.gmail.com
> > wrote:
>
> > Yun,
> > I developed part of the thiol-linkage parameter sets and a former
> graduate
> > student developed the initial parameter sets. We had planned to use
them
> > on
> > some synthetic glycans but that project didn't get funding. The
> > thiol-linkage parameters should be carefully employed and stringently
> > validated against experimental data since they were only developed on QM
> > models.
> >
> > I don't plan on developing new prep files for thiol-linkage glycans
> > however,
> > there is no reason you can't take the existing prep files and replace
the
> > linkage atom type "OS" and replace that with "SM". The only thing
you'll
> > have to do is re-develop charges which can be done with the RED server
if
> > you'd like. Be sure to generate Ensemble Averaged charges for each
> residue
> > that you use.
> >
> > Please let me know if you have any additional questions.
> >
> >
> > Matthew B. Tessier
> > Complex Carbohydrate Research Center
> > University of Georgia
> > +1 (706) 542-3508
> > mbt3911.uga.edu
> > matthew.tessier.gmail.com
> >
> >
> >
> > -----Original Message-----
> > From: Lachele Foley (Lists) [mailto:lf.list.gmail.com]
> > Sent: Thursday, September 15, 2011 2:40 PM
> > To: AMBER Mailing List
> > Subject: Re: [AMBER] about Glycam_06g.dat and thiol-glycosidic linkage
> >
> > We'll take a look at the paper.
> >
> > I can't make any promises about the prep file, but I'll see if anyone
> > has something in an advanced stage.
> >
> >
> > On Thu, Sep 15, 2011 at 2:07 PM, Yun Shi <yunshi09.gmail.com> wrote:
> > > Hi Lachele,
> > >
> > > Here is the link I used:
> > >
> > > http://pubs.rsc.org/en/content/articlelanding/2011/cp/c1cp20854c
> > >
> > > Should I expect a new version of prep file to include thiol glycosidic
> > > linkage in a few weeks? If not, I will try to do it myself.
> > >
> > > Thanks again.
> > > Yun
> > >
> > > On Thu, Sep 15, 2011 at 10:41 AM, Lachele Foley (Lists)
> > > <lf.list.gmail.com>wrote:
> > >
> > >> 1. Web of Science does not report finding an article containing the
> > >> words "Molecular dynamics studies of native and substituted
> > >> cyclodextrins in different media" in the title. Can you give a more
> > >> complete reference?
> > >>
> > >> 2-3. At the top of the parameter file, you will find atomic mass
> > >> definitions with useful comments. For example:
> > >>
> > >> N 14.01 sp2 N amide group
> > >> SM 32.06 sulfane carbohydrate linkage
> > >> (-CH2-S-CH2-)
> > >>
> > >> 2. Thiol linkages are still in the development phase. That's why
> > >> information for them occurs only in parameter file releases since the
> > >> GLYCAM06 paper. If you want to use them, you will have to generate
> > >> your own prep file entries. To do this, you need to generate
> > >> ensemble-averaged charges for the species you wish to model. You
> > >> should use the procedure as discussed in the GLYCAM06 paper. You
> > >> might be able to loosely base a prep entry on an existing one, but
> > >> that is best done only by someone with a lot of experience doing this
> > >> sort of modeling.
> > >>
> > >> :-) Lachele
> > >>
> > >>
> > >> On Thu, Sep 15, 2011 at 1:03 PM, Yun Shi <yunshi09.gmail.com> wrote:
> > >> > Hi Lachele,
> > >> >
> > >> > 1. I am confused after reading this paper -- Molecular dynamics
> > studies
> > >> of
> > >> > native and substituted cyclodextrins in different media: 1. Charge
> > >> > derivation and force field performances --, which compared the
> > >> performance
> > >> > of GLYCAM04 and GLYCAM06 on cyclodextrins.
> > >> >
> > >> > 2. I do want to a minimization with the force field and then
compare
> > it
> > >> with
> > >> > QM optimized structure. But I have some difficulty in building
> > topology
> > >> > file.
> > >> >
> > >> > Something like:
> > >> > .......
> > >> > SM 1.7210 0.2104 OPLS
> > >> > ......
> > >> > appears to the end of Glycam_06g.dat. Are these vdw parameters for
> > sulfur
> > >> > atom in something like a -CH2-S-CH2- environment?
> > >> >
> > >> > No residue as "SME" is defined in prep file, so I have to construct
> a
> > >> > monosaccharide like { OME 0hA } first, and then change the O atom
in
> > OME
> > >> to
> > >> > S, and SM atom type?
> > >> >
> > >> > Then what atomic charges should be assigned to the CG and SM atoms
> in
> > >> 'SME'?
> > >> > Should the anomeric carbon CG have a different atomic charge as
> well?
> > >> >
> > >> > 3. For the N -CG-CG-SM sequence, the N should correspond to an
amide
> > >> > nitrogen attached to C2?
> > >> >
> > >> > Thanks for the reply.
> > >> > Yun
> > >> >
> > >> >
> > >> > On Thu, Sep 15, 2011 at 8:32 AM, Lachele Foley (Lists) <
> > >> lf.list.gmail.com>wrote:
> > >> >
> > >> >> 1. GLYCAM06 should be used instead of GLYCAM04. The point of
> > >> >> GLYCAM06 is that it does a better job. Like I said in the other
> > >> >> email, if you find any place where 04 is significantly better than
> > 06,
> > >> >> please let us know. It should work well with other good protein
> > force
> > >> >> fields. The basic idea, and there might be times when this is not
> so
> > >> >> valid, is that if you make a really good, general force field for
> > >> >> carbohydrates and a really good, general force field for proteins,
> > >> >> they should, when used in concert, do a pretty good job modeling
> what
> > >> >> proteins and carbohydrates do together.
> > >> >>
> > >> >> 2. Any angle or bond length observed in a molecule is the
> > combination
> > >> >> of many different parameters. For example, if I build and
minimize
> > >> >> N-acetyl neuraminate { ROH 0SA }, I find that the bond length
> between
> > >> >> the ring oxygen and the anomeric carbon is 1.44. Their atom types
> > are
> > >> >> OY and CY, respectively. The parameter file, however, gives the
> > >> >> equilibrium bond length for those two types as being 1.410. The
> > >> >> difference is because all the parts of the molecule, all the
> > different
> > >> >> angle, bond and torsion parameters, push and pull on each other.
> > >> >> Certain aspects of any structure, therefore, are expected to
> deviate
> > >> >> from the values found in the parameter file. So, you can't look
at
> a
> > >> >> QM bond length or angle from within an entire molecule and expect
> it
> > >> >> to be exactly the same as any individual parameter. Conversely,
> you
> > >> >> cannot, with great precision, predict the bond lengths or angles
in
> a
> > >> >> molecule based on individual values in the parameter file. If you
> > >> >> want to know the final values in a molecule, you have to build the
> > >> >> molecule and, at the very least, minimize it using the force
field.
> > >> >> All of our parameters are taken from small molecule analogs. This
> is
> > >> >> a design philosophy that seems to work well and is generally not a
> > >> >> cause for concern.
> > >> >>
> > >> >> 3. It is, indeed, a typo. It should be:
> > >> >>
> > >> >> N -CG-CG-SM 1 0.45 0.0 -3.
> SCEE=1.0
> > >> >> SCNB=1.0 Thiol-linkages
> > >> >>
> > >> >> I will get corrected versions up on our site by the end of the
day.
> > >> >> Type "N" is defined at the top of the file.
> > >> >>
> > >> >> :-) Lachele
> > >> >>
> > >> >>
> > >> >> On Wed, Sep 14, 2011 at 6:53 PM, Yun Shi <yunshi09.gmail.com>
> wrote:
> > >> >> > Hi there,
> > >> >> >
> > >> >> > 1.
> > >> >> > In the end of this file, it notes that ".....in conjunction with
> > >> Parm94
> > >> >> > without introducing any conflict.....". So what about in
> > combination
> > >> with
> > >> >> > parm99 force field parameter set, would there be any foreseeable
> > >> >> problems?
> > >> >> >
> > >> >> > It seems many people are still using Glycam_04 series parameter
> set
> > >> for
> > >> >> > carbohydrate. I wonder if GLYCAM06 parameter sets are not as
good
> > as
> > >> >> > GLYCAM04 in terms of simulating pure carbohydrates?
> > >> >> >
> > >> >> >
> > >> >> > 2.
> > >> >> > In Glycam_06_alldocs.txt, it says that for Glycam_06g.dat,
> "Changes
> > >> >> include
> > >> >> > adding a comprehensive thiol-linkage set for glycosidic linkages
> > >> formed
> > >> >> by a
> > >> >> > Sulfur". So I looked into it, and found some differences between
> > the
> > >> >> ideal
> > >> >> > bond and angle values in the data set and those calculated for a
> > >> sample
> > >> >> > molecule from QM.
> > >> >> >
> > >> >> > I constructed a Met-S-alpha-L-Rhamose molecule, then did three
QM
> > >> >> geometry
> > >> >> > optimizations using HF 6-31G*, starting from three different
> > >> >> conformations.
> > >> >> > All three optimized structure showed a CG-SM bond length of
about
> > 1.83
> > >> >> > angstrom, while the value is 1.81 angstrom in Glycam_06g.dat.
And
> > most
> > >> >> > angles involved in the thiol-glycosidic linkage are off the
> values
> > in
> > >> >> > Glycam_06g.dat by 1 to 10 degrees.
> > >> >> >
> > >> >> > I understand angle parameters for i.e. CG-SM-CG are taken from
> > >> cysteine.
> > >> >> So
> > >> >> > I wonder if it's reasonable to still go head with what are
> > >> Glycam_06g.dat
> > >> >> if
> > >> >> > I am simulating a ligand containing this -S-alpha-L-Rham- part?
> > >> >> >
> > >> >> >
> > >> >> > 3.
> > >> >> > I also found something like:
> > >> >> >
> > >> >> > NH-CG-CG-SM 1 0.45 0.0 -3.
> > SCEE=1.0
> > >> >> > SCNB=1.0 Thiol-linkages
> > >> >> >
> > >> >> > in Glycam_06g.dat. Is this 'NH' a typo? If it is, what kind of
> > >> nitrogen
> > >> >> atom
> > >> >> > is it referring to? Like the nitrogen attached to C2 in a
> > carbohydrate
> > >> >> > derivative?
> > >> >> >
> > >> >> > Thanks for any advice.
> > >> >> >
> > >> >> > Regards,
> > >> >> >
> > >> >> > Yun
> > >> >> > _______________________________________________
> > >> >> > AMBER mailing list
> > >> >> > AMBER.ambermd.org
> > >> >> > http://lists.ambermd.org/mailman/listinfo/amber
> > >> >> >
> > >> >>
> > >> >>
> > >> >>
> > >> >> --
> > >> >> :-) Lachele
> > >> >> Lachele Foley
> > >> >> CCRC/UGA
> > >> >> Athens, GA USA
> > >> >>
> > >> >> _______________________________________________
> > >> >> 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
> > >> >
> > >>
> > >>
> > >>
> > >> --
> > >> :-) Lachele
> > >> Lachele Foley
> > >> CCRC/UGA
> > >> Athens, GA USA
> > >>
> > >> _______________________________________________
> > >> 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
> > >
> >
> >
> >
> > --
> > :-) Lachele
> > Lachele Foley
> > CCRC/UGA
> > Athens, GA USA
> >
> > _______________________________________________
> > 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
> >
> _______________________________________________
> 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
>
_______________________________________________
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 Wed Sep 21 2011 - 07:00:04 PDT
