I can't tell if my test results are good enough or not.
I have installed antechamber-1.27 to try to compute AMBER compatible
charge models for a number of small molecules found in the PDB. I am
using the version of mopac provided. I get some possible test
failures on all three test systems used by "make test". These are
primarily small differences in charges (on the order of +/- 0.02 or
less, although some are as large as 0.07, which makes me
uncomfortable) but there are also two different assignments of
chemistry -- in the NAD example, four of the P=O bonds change from
double to single ("2" to "1"), and the bond type of the C15-N16 bond
changes from "am" to "1".
Since I have not tried these calculations before, I don't know if
these levels of difference are expected or not.
I am using either g77 3.4.3 or gfortran 4.2.2 on Mac OSX 10.4.11,
with very similar but not identical results for the charges and the
same result for the changes in bond types.
I can send the actual diffs if they are needed.
Best regards,
Lynn Ten Eyck
Lynn F. Ten Eyck, E. T. S. Walton Visiting Professor
UCD Conway Institute Office: S055
Conway
University College Dublin Tel. +353 1
716-6845
Belfield, Dublin 4, Ireland email:
Lynn.TenEyck.ucd.ie
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber.scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
Received on Sun Feb 10 2008 - 06:07:34 PST
