AMBER Archive (2009)

Subject: [AMBER] update on am1-bcc charge problem in antechamber

From: case (
Date: Fri Aug 21 2009 - 11:32:30 CDT

This is an update to David Mobely's post about am1-bcc charges for

and followed up by Kevin Gilbert:

I have posted Junmei's bugfix to antechamber:

This changes how NO2 and SO2 groups are handled. If people know of other
places where there are problems with antechamber's assignments, we will
examine those as well.

With the above bugfix, antechamber and openeye agree on the am1-bcc
charges to within 0.001 charge unit for the 2-nitrothiophene example --
this probably just has to do with optimization termination cutoffs. Both
antechamber and openeye disagree with the values Chris Bayly assigned.
This may involve charge equivalence: the Bayly charges are the same for O1
and O2, whereas Amber and openeye have two different charges. We probably
need to nail down the algorithm that is used to assign equivalences; this
presumably involves some evaluation of the bond order or torsion barrier
for the ring-N bond. These points underscore the desirability of having a
database of "correct" am1-bcc charges, as pointed out earlier.

So, the bugfix addresses the big discrepancy for nitro groups, but leaves open
the question of smaller discrepancies with the Bayly charges, the question of
atom equivalences in charge assignment, and other places where antechamber may
be making the wrong assignments.

Thanks again to David and Kevin for their contributions here.


AMBER mailing list