AMBER Archive (2009)

Subject: [AMBER] Antechamber am1bcc questions

Date: Mon Jun 15 2009 - 08:04:19 CDT

  I have been asked to implement gaff in my open source mm engine so I have been
working on code to generate am1bcc charges based on the paper by Bayly et al.
Everything works fine, is the same as the output from Antechamber, except for
R-SO2-OR. So I ran the piroxicam example from the Bayly paper in Antechamber and
found some problems. For the SO2 group in piroxicam - Bayly uses a bond order
of 2 for S=O while antechamber uses a bond order of 9. Bond order of 9 is for
delocalized bonds such as nitro and carboxy and would probably be a good choice
for R-SO3- , but is probably not a good choice for this case. Also the nitrogen
adjacent to the SO2 group is typed as 21 in Bayly but as 23 by antechamber, with
type 23 being for a delocalized nitrogen lone pair. In this case I think the
antechamber choice is better since the nitrogen is also adjacent to a double
bond and is probably closer to an enamine nitrogen than a simple amine.

  Has anyone noted these problems?
  Has there been an update to the Bayly atom and bond typing paper?
  Since Antechamber clearly conflicts with the published results which one
should we use?


Kevin Gilbert
Serena Software

AMBER mailing list