From:	Patrick Bultinck <Patrick.Bultinck-0at0-rug.ac.be>
Date:	Tue, 16 Nov 1993 19:02:22 +0100 (MET)
Subject:	Summary / MM parameters macrocycles???

Hi netters,

About a week ago I wrote the message beneath. I would like to thank
everybody who gave his opinion, advice or comment to the questions I
asked. Also many thanks for the good-luck wishes...
Please find a summary of the answers below.

MY MESSAGE :
Dear Netters,

On Nov. 24th. I will have to pass a test to see if I get a
scholar-ship for the next year of  my doctoral research. I am now going to
start performing ab initio calculations involving complexes formed between
some alkaline-earth metal ions and some crown structures and macrocycles.
In my project paper I referred to molecular mechanics as the most widely
used (but non-quantumchemical) method to computationaly study these
compounds. I have found some articles in which was stated that a major
problem with MM is that the method is only as good as the used parameters.
There appear to be some problems obtaining good values for these
parameters. As the examination jury are probably going to ask me why this
is such a problem I would like to know if someone can help me with a
reference to some article that explaind the trouble or if someone knows a
good textbook.
Off course I would also be grateful if someone just gives his opinion why
this constitutes a problem !
Summary : hopefully before the end of the week

***********************Patrick Bultinck**********************************
************Lab. for General and Inorganic Chemistry*********************
**********************University of Ghent********************************
****************************Belgium**************************************
**************E-mail : Patrick.Bultinck[ AT ]rug.ac.be************************

But come on Heisenberg you must be sure about something.........


RECEIVED MESSAGES:

From: "John M. Simmie, University College, Galway, Ireland"
 <0001877S -AatT- bodkin.ucg.ie>
Subject: mm parameters
To: patrick.bultinck -AatT- rug.ac.be

Patrick:
Molecule A consists of atom X bonded to Y, Z,   etc
Molecule B consists of atom X bonded to Y, Z,   etc

deduce the parameters X(Y)(Z).. from fitting the experimental data for A
now attempt to predict data for B from parameters

BUT A is not = B therefore no good reason why prediction should be exact
(close maybe as you expand the number of parameters but this self-defeating if
taken too far)

Regards, John

               -------------------------------------------------
                              Dr. John M. Simmie
               Chemistry Department             Roinn na Ceimice
               University College         Colaiste na hOllscoile
               Galway                                   Gaillimh
               Ireland                                      Eire
               Fax: (+353)-91-25700         Fsn:(+353)-91-750388
               -------------------------------------------------
                            XRASIMMIE %-% at %-% BODKIN.UCG.IE


From: Tilman Brodmeier 
To: Patrick.Bultinck-0at0-rug.ac.be
Subject: RE >> MM parameters/macrocycles ???

Hi,
Have a look at the paper of my collegue:

M. Badertscher et al., J. Comput. Chem. Vol 11, No 7, 819-28 (1990)

Hope this can help you, bye
Tilman

From: young <-at-> slater.cem.msu.edu (Dave Young)
To: Patrick.Bultinck "-at-" rug.ac.be
Subject: MM parameters


Hi,

	I thought I would point out some things about molecular mechanics
that will seem obvious once I explain them but are often overlooked.

	What molecular mechanics does is to use an analytic form,
such as a Morse potential for each of the forces acting on the atoms
in a molecule.  The constants within these equations are obtained
from experimental or (less often) ab initio data.  There are several
direct consequences of this method.

	First:  The analytic form being used may not describe the
actual shape of the potential.  There are a number of analytic forms
being used.  A Morse or Leonard-Jones potential has a fair chance
of describing dissociation, but they are not exactly correct
for all molecules.  Some force fields even use a harmonic potential
that won't dissociate at all, but they get reasonable equlibrium
geometries.  For an extreme case, look up the dissociation curve
for the Cu2 dimer It has extra maxima and minima in it that none of
these analytic expressions can touch.

	Second:  Even if you adequately describe the behavior of
an atom in one environment, it may behave differently in another
environment.  Thus to do really well you have to make new force field
parameters fitted to experiment for every simulation you run.
If you are willing to just accept the parameters in a known force field
you are taking a good average behavior, but have no guarantees on how
well it will work on your system.  For this reason, you must be very
careful in how much you trust the results from all-purpose force fields
such as MM2, MM3 or Dreiding.  This is also why there are specific
force fields such as the "protein only" force fields.

	Third:  The behavior of the chemistry is modeled based on
which atoms are bonding together.  I don't know of any force field
taking second nearest neighbor interactions really well into account.
For this reason, the difference in chemistry between a benzene ring
and a benzene ring with an electron withdrawing group attached are
too subtle for any standard force field.  You could try doing this sort
of thing by having two complete sets of parameters for these
two types of benzene rings, but it would be really hard to get exactly
right.

	Now, the way that you usually use mechanics is as follows.
First look at the stuff I said above and ask yourself if it has a chance
of working for the system you are interested in.  Second, look through
the literature for mechanics studies of systems similar to your own
and see which force fields they used, whether they came up with a few
new parameters for that specific case, and to what extent it was
qualitative or quantitative when compared to experiment.  If you need
to fit your own parameters, you will either have to be really careful
and follow closely how people justify their choices or work with someone
who does this sort of thing.  Present your work in guarded speach such as
"We realize that there are appoximations in the force field approach,
but this force field has a history of correctly giving the qualitative
results we are interested in with systems very similar to ours."

	The mechanics/dynamics force field approach to chemistry can be
a really great way to model chemistry as long as you always ask yourself
if the method and results are chemically sensible and don't trust them
for more than they can give.

	Good Luck


				Dave Young
				young at.at slater.cem.msu.edu
				youngdc at.at msucem




From: alex &$at$& tammy.harvard.edu (Alex Mackerell)
To: Patrick.Bultinck(-(at)-)rug.ac.be
Subject: parameters


Patrick,

   Congratulations, you have achieved a level of enlightment concerning
the inherent limitations in empirical force field calculations which
a majority of individuals using empirical force field calculations never
achieve.

Basically, the simplicity of the mathematical model used in empirical
force field calculations leads to the limitations that the approach
is only as good as the parameters employed.  This simplicity is required
due to the large size of the systems being studied via empirical
approaches (thousands of atoms). In order to properly parameterize systems
extensive amounts of experimental and ab initio data must be taken into
account to insure that the parameters are not biased.  As computer
resources increase and the amount of experimental data to use as
the basis for the development of parameters increases the quality of
the parameters coninues to improve.  Thus, empirical force field parameters
will always contain room for improvement and, more importantly, when
performing and analyzing results from empirical force field calculations
the potential for the parameters to bias those results must always be
taken into account.

It should be remembered that all theoretical approaches for the study
of chemical systems employ models which attempt to reproduce reality.
None of those models are reality.  Even molecular
orbital calculations, which use a much more detailed model as compared
to empirical force field calculations, also
have limitations concerning the accuracy of the obtained results.

I have spent the last 6 years optimizing the CHARMM22 all-hydrogen
empirical force field parameters for proteins, nucleic acids and
lipids.  The end is not in sight.

Good luck with your examination.

Alex MacKerell


From: burkhart %! at !% goodyear.com (Craig W. Burkhart)
To: Patrick Bultinck 
Subject: Re:  MM parameters/macrocycles ???

Patrick:

There are several reasons why you should always look at a molecular
mechanics force field with a great deal of scepticism. Among them
are:

1. Force fields are usually tailored, through the use of its own database
   of structures, for a particular application in a particular range of
   temperature.

2. The form of the potential energy function is at best a crude
   approximation of the underlying quantum mechanical effects
   (e.g., polarizability). So whether you are using a standard Lennard-Jones
   or Buckingham function, be certain that it too is a MODEL of
   reality.

3. In dealing with special energy functions, such as hyperconjugation
   or hydrogen bonding, the form of the functions (Fourier parameterization,
   angle-dependent LJ functions, for example) the data used to make the
   coefficients are difficult to test for "goodliness".

There are many more reasons to be sceptical, of course, but in my opinion
these are the primary culprits. Be especially cognizant of Item 1. In your
studies you should ALWAYS use a parameter set that has been "tuned" to
your problem. DO NOT use a parameter set based on proteins for work with
the thermodynamics of hydrocarbons. If you cannot readily find a force
field suitable for what you want to accomplish, you will have to derive
one for yourself. In your case you are interested in alkaline-earth
metal complexes. Note that ionic species are difficult to parameterize,
since the form of the potential energy function is different for each
family. For example, the potential energy form is different for alkaline
earth ions and transition metal ions. In the case of ionic species, you
should also be sensitive to the fact that you will have to use a RELIABLE
means of computing the charges of the species if it has covalent character
(as in the case of some organolithium molecules). Just using MNDO is NOT
enough.

As for a reference on the usage and form of various potential functions,
I can think of no better reference than this classic work:

J.O. Hirschfelder, C.F. Curtiss and R.B. Byrd, THE MOLECULAR THEORY OF
GASES AND LIQUIDS, Wiley, 1964.

My copy has a green binder--thus, I call it the "green bible". It should
be the first stop of anyone interested in the statistical mechanics of
molecular systems. In Chapter 8, you will find the answers to why the
temperature range is important as well. I have been in the field for
about 13 years, and this book has never been far from my side.

Good luck on your examinations. If I can be of further service, contact
me at your convenience.

--------------------------------------------------------------------------
Craig W. Burkhart, Ph.D                    Senior Research Chemist
E-mail: cburkhart #*at*# goodyear.com             The Goodyear Tire & Rubber Co.
Fone:   216.796.4431                       Akron Technical Center
Fax:    216.796.3947                       Akron, OH   44309-3531
--------------------------------------------------------------------------
"For a successful technology, reality must take precedence over
 public relations, for Nature cannot be fooled." - Feynman
--------------------------------------------------------------------------


From: stevens &$at$& rainbow.uchicago.edu (Jonathan Stevens)
Apparently-To: Patrick.Bultinck ^at^ rug.ac.be

Patrick,
     Here is one of those opinions you asked for.
The problem is probably one of transferability. MM is like semiempirical
theory in that the parameters used is the calculations are obtained by
"fitting" them so that calculations on molecules that are known
experimentally are made to reproduce the experimental data. The parameters
thus obtained are then "transferred" to systems similar (but of course not
identical) to the system from which the parameters were obtained. Since the
new system is similar but not identical to the old, the parameters may or
may not be good for accurate calculations of that system. Discussions of
this phenomenon can probably be found in any book about semiempirical
theory (or in your case MM); this is where you would have to go for more
details. I know of no such book about MM myself.
     Hope this helps.
                                      Jon Stevens

"Your brush with greatness is over." -"Flyin'" Brian Pillman


From: sling "-at-" silver.ucs.indiana.edu
To: Patrick.Bultinck-: at :-rug.ac.be
Subject: Re:  MM parameters/macrocycles ???


In principle, everything in the natural world should be treated quantum
mechanically (see, e.g. E.H. Wichmann, Quantum Physics, Berkeley Physics
Courses, Vol. 4, McGraw Hill, 1971), this includes both microscopic and
macroscopic objects, from atoms and molecules to pendulums and clocks.
However, the correspondence principle telles us that the quantum treat-
ment to macroscopic objects would give us exatly the same results as that
of classical mechanics.  (Do not be bothered by quantum chaos here, at least
for the time being  Molecules are small enough that they should be treated
quantum mechanically, but the present technology still presents much difficulties
in so doing, therefore the molecular mechanics which is basically applications
of classical mechanics.  To some extent, molecular mechanics might be able
to account for some phenomena that are classical in nature - treating atoms
like little balls or point masses and ignoring their wave nature, but it is
generally speaking not a valid approach in principle for atoms are waves as
well as particles, it is the interference of the waves that form and break
the chemical bonds, for example.  Consequently some experimental parameters
might be set in to guide the computations, some adjustable paramenters might
be used to "variationally" match the experiments, and so on.  These are
natural extensions of the incompetence of the principle matter behind:
classical mechanics.  I am not saying that there is anything wrong so doing,
but definitely like anything else that has a label of time value in science
and technology, molecular mechanics will not be the ultimate theory for
molecules because of its starting point.


From: susan jackels 
Subject: Re: MM parameters/macrocycles ???
To: Patrick Bultinck 

Dear Patrick,

There's an article by Benjamin Hay in Chem. Revs. (1993) on use of MM
with crown ethers and alkali metals. Hope you find it useful.

Susan C. Jackels
Department of Chemistry
Wake Forest University
Winston-Salem, NC  27109
Phone: (919)759-5514     FAX: (919)759-4656
Internet: sjackels #*at*# ac.wfunet.wfu.edu

On sabbatical for 1993-1994 at:
Department of Medicinal Chemistry
308 Harvard Street S.E.
University of Minnesota
Minneapolis, MN 55455
Phone: (612)626-4429




To: Patrick.Bultinck : at : rug.ac.be
From: dwhite (+ at +) assistant.beckman.uiuc.edu
Subject: MM Parameters/Macrocycles

>There appear to be some problems obtaining good values for these
>parameters. As the examination jury are probably going to ask me why this
>is such a problem I would like to know if someone can help me with a
>reference to some article that explaind the trouble or if someone knows a
>good textbook.
>Off course I would also be grateful if someone just gives his opinion why
>this constitutes a problem !

For a godd philosophical type introduction to MM, there is an ACS monograf
available (ACS monograf # 177, by Ulirch Burkert and Norman L Allinger).  I
would say the essential problem with MM and transition metals is the wide
variety of geometries and oxidation states available compared to the
organic family.  In part this problem has been addressed by Rappe with the
introduction of the Universal Force Field and one of his papers has an
excellent introduction to MM and transition metal chemistry (Rappe, A.K.;
Colwell, K.S.; Casewit, C.J. Inorg. Chem. 1993, 32, 3438).  As far as the
specific issue of chelkates and macrocycles goes, Rob Hancock has done a
lot of work on this particular system.  His papers are referenced in
Rappe's (one is not: Hancock, R.D. Prog. Inorg. Chem. 1989, 37, 187).   I
know Rob is on meail, but do not have his address, at the moment he is a
visiting professor at Texas A&M.  One of his PhD graduates, Peter Wade, may
well have some valuable insights for you, Peter can be contacted at
pwade "-at-" nuustak.csir.co.za.  I hope this helps.

Good luck for the oral.

Dave








Similar Messages
10/13/1998:  Parameterization - summary
08/01/1996:  Re: CCL:M:Heat of formation calculation using MOPAC.
03/02/1992:  Oh, boy, oh, boy!  Real scientific controversy!
05/08/1995:  AM1 vs. PM3
10/01/1993:  torsion of conjugated systems -- summary
04/12/1994:  AM1 vs. PM3
11/02/1995:  summary AM1 vs PM3
04/18/1994:  Semiempirical parameterization yet again...
06/08/1993:  undergrad computational chem
11/26/1996:  summary: pi-pi and tm-params with semiemp methods


Raw Message Text