[AMBER-Developers] CHCl3 parameters

From: Brent Krueger <kruegerb.hope.edu>
Date: Mon, 18 Jun 2012 15:33:06 -0400

AMBER experts:

There was a brief discussion on chloroform parameters some time back, but
to my knowledge no one really came to any conclusions. I am starting some
chcl3 simulations now and so I've had to dig through some older papers and
am confused about the origin of the parameters given in the frcmod.chcl3
file that ships with AMBER11. (I've not checked amber12, but this frcmod
file has not changed in many years, so I suspect the story is the same in
12.) Anyway, I found the history of these parameters to be confusing, so I
thought I would summarize what I found here. More importantly, I am
ultimately left questioning whether the frcmod.chcl3 parameters are
correct, so I'm hoping that someone who knows more than me can help out.

It looks to me that the parameters provided in the frcmod.chcl3 file are
those used by Piotr and Jim in the Cieplak, Caldwell, and Kollman paper
from J Comput Chem 2001, 22, 1048. In that paper, they mention using chcl3
parameters from Fox & Kollman (J Phys Chem B, 1998, 102, 8070) except that
they modified the epsilon parameter to give better liquid properties.
 However, this does not appear to be quite correct. The Cieplak parameters
match those given by Fox for a RIGID chcl3 model, whereas I believe Cieplak
et al were using a FLEXIBLE chcl3 model. Certainly, in current AMBER if
one runs a standard simulation, the chcl3 is flexible. Note that we are
talking only about the nonbonded parameters of the CL. All of the bond
and angle parameters in the frcmod.chcl3 match those given by Fox -- and
they are the same whether the chcl3 model is rigid or flexible. To add to
the confusion, a 1997 paper by Eksterowicz, Miller, and Kollman utilizes a
flexible chcl3 model, but presents parameters that match the Fox rigid
model. The CL nonbonded parameters from these papers are summarized like

R* and epsilon for the CL in chloroform
Eksterowicz (1997) FLEX 2.0000 0.255 (based on older Fox paper that
modeled rigid chcl3)
Fox (1998) RIGID 2.0000 0.255
Fox (1998) FLEX 1.9480 0.265
Cieplak (2001) FLEX? 2.0000 0.325

My guess at what happened is that the 2.0000 / 0.255 parameters for rigid
chcl3 had existed for several years. Eksterowicz (and probably others)
used them with flexible chcl3 because no other parameters existed at the
time. Fox then found that the parameters could be improved for flexible
chcl3 and published those as a side point in his 1998 JPCB paper. Then
maybe that information was lost and when Piotr and Jim started working with
chcl3 they went back to the rigid parameters and rediscovered that they
weren't ideal for flexible chcl3. They then developed their own improved
parameters by varying only epsilon rather than both sigma and epsilon. And
so today we now have two different parameter sets to describe flexible
1.9480 / 0.265 by Fox & Kollman
2.0000 / 0.325 by Cieplak, Caldwell, and Kollman.

What is not clear to me is which set of parameters is preferred. Perhaps
Piotr or someone else involved remembers some details? Or perhaps someone
else has some input here?

It is certainly the case that the Fox & Kollman paper has a more complete
description of what was done in the parameterization, describing that they
attempted to reproduce experimental density and heat of vaporization. The
Cieplak, Caldwell, and Kollman paper simply has a reference to the Fox
paper along with the statement that they "found better liquid properties
with epsilon = 0.325 kcal/mol, with all other parameters the same." This
leads me to believe that the Fox parameters were more carefully determined.
 However, I tend to trust things that Piotr and Jim worked on, so I hate to
just ignore that.

In a side note -- the atom types specified in the CHCL3BOX file include CZ
for the carbon and HZ for the hydrogen. Since CZ and HZ do not exist in
parm94, this was a sensible choice for a residue with unique parameters.
 However since they do exist in parm99, it means that anyone who uses
parm99 and then reads in frcmod.chcl3 will be overwriting the standard CZ
and HZ with parameters that are unique to chloroform -- definitely not good
replacements for sp-hybridized carbon. Probably the atom types could be
changed in the CHCL3BOX and the frcmod.chcl3?


Brent P. Krueger.....................phone:   616 395 7629
Associate Professor................fax:       616 395 7118
Hope College..........................Schaap Hall 2120
Department of Chemistry
Holland, MI     49423
AMBER-Developers mailing list
Received on Mon Jun 18 2012 - 13:00:03 PDT
Custom Search