[gmx-developers] Validated OPLS-AA files

[Shirts, Michael (mrs5pt)]

Hi, all-

So, I noticed an email to the user list about OPLS parameters, which
reminded me.  A few months back, I did an extensive comparison of Gromacs
OPLS-AA with Macromodel and Desmond implementations of OPLS-AA, which are
about the closest to a reference implementation there is: in some sense BOSS
could be considered a better reference, but there's no versioning there. And
a lot more people are running Schrodinger/Desmond codes now than BOSS.

>From that comparison, I put together a version of OPLS-AA in gromacs that
-almost- exactly replicates the Macromodel/Desmond energies.* Basically, the
differences between the 3.3 OPLS and Macromodel/Desmond that some charges
for alternate protonation states for some residues, as well as a few
torsions.

Right now, all my files  all in the old ff format, and I don't have the time
to reformat it; but I'd be willing to turn over all the information (along
with all the validation documentation) for somebody to get it working for
4.5.  It could be that several of these changes have been made in 4.0, since
I was working with 3.3 at the time; I haven't checked yet.

*The one exception is that because all three programs
(Macromodel/Desmond/Gromacas) have slightly different ways of handling the
atom ordering for impropers, different ways of specifying the impropers for
the same molecule will lead to slightly different improper energies among
everything.  And there is no standard for specifying the improper atoms
order.  But at the minimum of the impropers, everything is exactly equal up
to rounding in conversion factors.

Best,
~~~~~~~~~~~~
Michael Shirts
Assistant Professor
Department of Chemical Engineering
University of Virginia
michael.shirts at virginia.edu
(434)-243-1821