Hi Conor,
I have never looked into simulating a crystallization process, and I am
not familiar with the literature concerning this. I can imagine that the
nonbonded forces will be very important, both Lennard-Jones and Coulombic.
After that I would assume that proper molecular structure would play some
role. Perhaps a good literature search will help to reveal some information
on this.
I am personally motivated to try and capture the forces and conformations
of a chemical/biological model as best as I can. If I know there are
questionable conformations, found by comparison to experiments or reliable
QM data, then I try to improve upon the force field. Then when you do run
your simulation and you are analysing your results, you have a bit more
confidence that the conformations seen in your simulation are reasonable.
Otherwise you will execute your simulations, do the analysis, and then have
to consider the impact that errors in the dihedral parameters might have.
As I tell students, we build artificial computational models, for which we
have made some assumptions and approximations, which in turn leads to some
introduction of errors into the the resulting observables. To be able to
best analyse your computational results, you need to know what assumptions
have been made in your model building, and what errors have been
introduced. Sometimes these errors are not significant to the observables
your are looking at, sometime they are.
Hope this is helpful, even if a bit philosophical.
Bests,
Karl
On Fri, May 6, 2016 at 1:13 AM, conor parks <coparks2012.gmail.com> wrote:
> Karl,
>
> Thank you so much for looking into this so thoroughly. I don't honestly
> know if the dihedral effect is important. The goal of my work is to
> simulate crystallization of acetaminophen form I and form III. For self
> assembly of organic molecules, or solid state crystal crystal calculations,
> much care is typically paid towards the selection of the partial atomic
> charges. However, the dihedral angle importance isn't as clear to me. Do
> you have any thoughts here? My experience rests solely in MD, and I am not
> familiar with the various levels of DFT theory you are referencing.
> Although, the differences between the MD and DFT dihedral energy graphs is
> obvious.
>
> Can someone answer how to evaluate the dihedral energies of an angle that
> has multiple parameters from the gaff.txt file? I am implementing this in a
> group MD code, and want to make sure the procedure is right. I am currently
> assuming I just evaluate the dihedral energy for each parameter set, and
> sum over the multiple parameter sets.
>
> Regards,
>
> Conor Parks
> B.S.E in Chemical Engineering, University of Michigan, 2012
> PhD candidate in Chemical Engineering, Purdue University
>
> On Wed, May 4, 2016 at 4:28 AM, Karl Kirschner <k.n.kirschner.gmail.com>
> wrote:
>
> > Hello Conor,
> >
> > I got curious about your molecule and took a look at it. I did some
> quick
> > HF/6-31G(d) calculations and found two minima that were close in energy -
> > one that is the conformation you sent out, and a second lower energy (by
> > 0.43 kcal/mol) conformation where the amide group functional group is
> > rotated 180 degrees (see attached figure that contains the AM1-BCC
> charges
> > for this conformation). I verified that they are minima on the potential
> > energy (PE) surface by doing frequency calculations. I also did two sets
> > of torsion rotations, one for each functional group of the phenyl ring. I
> > then compared how the gaff parameters perform using the AM1-BCC charge
> set
> > (see attached PE figures). The rotation of the hydroxyl group looks
> decent,
> > in my opinion. However, the rotation of the amide functional group
> deviates
> > significantly from the HF/6-31G(d) results. Around both HF/6-31G(d)
> minima
> > (0 and 180 degrees), the molecular mechanics (MM) calculations predict a
> > double well potentials.
> >
> > I would say these are preliminary results. To really verify the QM
> > surface is reliable, I would suggest doing the calculations at a more
> > rigorous theory level (e.g. MP2/aug-cc-pVDZ or MP2/cc-pVTZ). Right now, I
> > don't truly know if the HF/6-31G(d) or MM results are correctly
> describing
> > the shape of the PE curves. (However, from experience I believe that the
> QM
> > results are correct.)
> >
> > The question now becomes is how important are these small
> conformational
> > changes for your investigation. If they are very important, then you
> might
> > have to do some parameter optimization.
> >
> > Hope this helps in thinking about your model.
> >
> > Bests,
> > Karl
> >
> > On Tue, May 3, 2016 at 10:13 PM, conor parks <coparks2012.gmail.com>
> > wrote:
> >
> > > David,
> > >
> > > thank you very much for providing that link. That was very helpful.
> That
> > > clears up the negative sign issue for me. Thanks! Was my first
> question,
> > > correct? I.E. the dihedral potential and forces get evaluated for each
> > > parameter set associated with it, even when there are multiple, and the
> > > results are summed?
> > >
> > > Regards,
> > >
> > > Conor Parks
> > > B.S.E in Chemical Engineering, University of Michigan, 2012
> > > PhD candidate in Chemical Engineering, Purdue University
> > >
> > > On Tue, May 3, 2016 at 3:53 PM, David A Case <david.case.rutgers.edu>
> > > wrote:
> > >
> > > > On Tue, May 03, 2016, conor parks wrote:
> > > > >
> > > > > However, in my prmtop file (values given in first email in this
> > chain),
> > > > the
> > > > > periodicity values are given as 1 and 2 (no negative sign). What
> > > happened
> > > > > to the negative sign on these periodicity values in my prmtop file?
> > > >
> > > > Please see http://ambermd.org/formats.html#parm.dat. Negative
> values
> > > in a
> > > > parm.dat file have a special meaning.
> > > >
> > > > ...dac
> > > >
> > > >
> > > > _______________________________________________
> > > > 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
> > >
> >
> >
> >
> > --
> > Karl. N. Kirschner, Ph.D.
> > Research Associate
> > Bonn-Rhein-Sieg University of Applied Sciences
> > Grantham-Allee 20, 54757 Sankt Augustin, Germany
> >
> > _______________________________________________
> > 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
>
--
Karl. N. Kirschner, Ph.D.
Research Associate
Bonn-Rhein-Sieg University of Applied Sciences
Grantham-Allee 20, 54757 Sankt Augustin, Germany
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Sat May 07 2016 - 12:30:04 PDT
