Re: CCL:Gasteiger charges questions
This is a response mailed on behalf of Prof. Gasteiger
(gasteiger ^at^ chemie.uni-erlangen.de) himself:
Dear Mr. Leonard,
you might be surprised that Gasteiger is still a living person - and
still going strong. A quick internet search should have brought you to
our web site. Then you could have addressed me directly.
In any case, let me try to answer your questions:
The calculation of partial charges in organic molecules falls into two
parts: the calculation of sigma-bonded systems and that of pi-bonded
systems. There exist quite a few implementations of our approach to
charge calculation and I do not know how good a job other people -
including commercial companies - have done in reproducing our work. The
calculation of charges in sigma-bonded systems (PEOE) has been
explicitly published such that anybody that wanted to implement it
should have been able to do so correctly. The approach is still valid
and all parameters are correctly given in the paper.
The calculation of charges in conjugated pi-systems (PEPE), however, is
somehow more complicated. Here, you have to first calculate sigma
charges and then relax the pi electrons to the sigma charges and sigma
and pi electronegativities. Our first approach was presented at a
meeting in Croatia and the proceedings were published in Croatica
Chimica Acta (I agree that this is not our regular journal for
publishing our research!) We further modified our approach and then
published it in Angewandte Chemie (J.Gasteiger, H.Saller,
Angew.Chem.Intern.Ed.Engl. 24, 687-689 (1985)). This is only a
communication giving the essential ideas which makes it not
straightforward for people to implement our method. However, any charge
calculation in conjugated systems that only considers the sigma method
(and unfortunately many people only do this) must provide bad results.
It is true that the methods have been developed more than 20 years ago
and we had thought of supplementing our methods by quantum mechanical
approaches. However, with the advent of combinatorial chemistry people
have to deal with really large datasets - and we have used our charge
calculations on datasets with millions of structures! Thus, we have been
working with these methods over all the years and we have even
reimplemeted it recently in C++ as interest in fast calculation of
charge continues to go strong. Our newest approach is based on a Huckel
method that modifies the parameters based on the sigma charges.
This approach is contained in our program package PETRA for the
calculation of physicochemical effects in organic molecules. You can
access these calculations from our web site
(http://www2.chemie.uni-erlangen.de) and use it on molecules
of your
interest. There is also a manual of the PETRA package on our web site.
The PETRA package is also commercially distributed by the company
Molecular Networks (info ^at^ mol-net.de).
If you provide me with your mail address I send you a copy of the
Croat.Chim.Acta paper (of historical interest) and of the Angew. Chem.
paper.
I hope I have helped you with my comments,
regards
Johnny Gasteiger
--
Dr. Wolf-D. Ihlenfeldt
Computer Chemistry Center, Institute for Organic Chemistry,
University of Erlangen-Nuremberg
Naegelsbachstrasse 25, D-91052 Erlangen (Germany)
Tel (+49)-(0)9131-85-26579 Fax (+49)-(0)9131-85-26566
http://www2.ccc.uni-erlangen.de/wdi/