Similarity in Organic Chemistry -
A Summary of the Beilstein Institute Conference
Stephen R. Heller
USDA, ARS, NPS
Beltsville, MD 20705
As chemists increase their use of computer systems, a number of conferences have been established to meet
the needs of the community. The biannual International Conference on Computers in Chemical Research and
Education (ICCCRE) (1) held its 10th conference in Israel 1992 (2).
A more recently established conference is the
Beilstein Bozen (Bolzano) Meeting. This is also a biannual meeting and the third one in the series was held in May
1992. The papers from the 1992 Beilstein Bozen meeting are published in this issue of the Journal of Chemical
Information and Computer Sciences.
Bozen conferences have previously been held on the Estimation of Physical Properties of Organic
Compounds (3) and on Computer Reaction Management in Organic Chemistry (4). The 1992 Beilstein Bozen
Conference, attended by some 60 scientists from Europe, the USA, and Japan, took as its theme Similarity in
Organic Chemistry. This paper summarizes the 31 papers and poster talks presented at the meeting, 24 of which
are printed elsewhere in this issue.
Similarity in chemistry is a very venerable notion. The periodic table itself was developed from the
observed similarity in the behavior and properties of various elements and it stands as perhaps the most widely
known and visible example of similarity in chemistry. The last twenty years has seen a great deal of work on exact
matching of structures and substructures but now similarity, or inexact structure matching has recently been given
a new examination and a number of workers have been providing this area of science with a more rigorous
mathematical and conceptual foundation. Prominent among these efforts is the recent book by Johnson and
Maggiora (5) on the subject. In fact many of the presentations that were made at Bozen in 1992 and which are
printed in this issue of the Journal of Chemical Information and Computer Sciences represent continuing studies
by workers who, in 1988 and 1989 wrote chapters for Johnson and Maggiora's book.
A literature search of Chemical Abstracts for the term "similarity" produced the following results:
|CA Period||# Hits|
This is not of course a rigorous measure of the importance of this subject area, but it does suggest that there
is increasing interest on the part of the scientific community in the concept of similarity.
The areas of chemistry which were covered in the 1992 Beilstein Bozen meeting were quite varied and extensive. D. Rouvray provided an excellent introduction to and review of the overall area of similarity(6). A number of papers described different various methods for assessment or measurement of similarity. These included "quantum similarity", a method developed by Carbo and based upon quantum-mechanical priniciples7, and geometrical similarity, or shape analysis approaches described by Venkatachalam and by Mezey (8). The importance of chemical similarity in reactions was the focus of a number of papers. Lawson (9) describes the concept of similarity between reactions and its use in information classification. Ponec (10) discusses the use of reaction similarity to rationalize various mechanistic features of pericyclic reactions, and Fontainti reports on the application of genetic algorithms that can determine constitutional similarity. A program, "Beppe" which predicts the likely products of specific reactions was described by Sello (12), Grethe discussed the similarity searching capabilities of the MACCS-II software, Mark Johnson (13) reported on the use of "site environments" in substrates to distinguish between xenobiotic amines susceptible to N-oxidation and those which are instead demethylated, and Gladkova described a new retrieval language based upon bond counts in molecules and showed that it is useful in, inter alia, reaction indexing. Synthesis design and planning can be influenced positively by considerations of similarity. Thus Gasteiger (14) showed how quantitation of the electronic effects at reaction sites can be used to determine the similarity between reactions and Hanessian (15) described the use of his program "Chiron" to identify the similar architectural features common to different molecules and so predict biogenetic pathways. The application of similarity in structure and substructure searching was dealt with by a number of speakers. Willett (16) described the use of atom mapping as a means of three-dimensional similarity searching, Fisanick (17) reported on the work of the CAS group on the use of similarity searching to find "fuzzy" matches in a database and Peter Johnson (18) discussed his multiple largest common subgraph routine, which finds the largest substructure
common to several different molecules. Takahashi (19) covered the use of automatic similarity determination in QSAR, Razinger described programs for the exhaustive generation of the stereoisomers corresponding to a given structural formula and Marsili (20) provided some detail on his program ANALOGS, which generates sets of compounds related to a lead compound and useful in QSAR studies. Perry (21) reported on the use of molecular similarity in 3D searching, and Barnard (22) described his work on the clustering of chemical structures
on the basis of 2-D similarity measures. The use of similarity in QSAR was discussed by Judson (23) and Allan
(24) described a novel momentum-space approach to the similarity of electron densities and hence of molecules.
Randic (25) discussed the use of an orthogonalized extended list of descriptors as a measure of similarity,
but apart from this one paper there was less presented concerning the presentation of similarity measures than one
would have wished, although in three days it is hard to completely cover such a broad subject matter.
In the area of structure searching the value in similarity searching is that one is able to do a "loose" or
"fuzzy" search, as opposed to the usual "tight" search and we saw that having a good deal of slack in a search can
lead to some interesting search results. One very important area of structure similarity searching which was not
addressed was Markush searching (26). The related area of homology searching in libraries of DNA base pairs was
also not covered.
One session of the meeting was devoted to neural networks. This session began with a paper by Maggiora
(27) who gave an excellent introduction to the use of neural networks in some chemical problems, as well as the
use of artificial neural nets to model highly complex systems in the area of structure-activity research. Other papers
in this session included one by Clark on the use in neural networks of data developed by semiempirical MO theory
in comparison with physically observed data, and a description by Herges of the use of neural nets for
determination of aromatic substitution patterns from infrared spectra. Finally, Kvasnicka spoke on the application
of neural nets to prediction of CNMR shifts and meta-product distribution of nitration in a series of mono-aromatic
To paraphrase the old proverb: "Similarity is in the eyes of the beholders. As indicated by the varied
subjects that were touched upon during this workshop, there are many ways to view similarity. There is little
question that many new developments in applied organic chemistry, and especially pharmaceutical chemistry could
come from similarity-based approaches. If one knows enough about what the actual process or problem to be
solved (e.g., design of molecules that interact with a particular receptor site) an appropriate measure of similarity
can be chosen which allows one to work on solving the problem most quickly and efficiently.. One can easily
envision using an appropriate similarity measure to find bio-isoteres as new drug alternatives which would not
violate an existing patent.
It is a paradox that more precise tools such as high-quality and high-resolution instruments and computers often lead us to more imprecision in science, particularly in the area of biological systems. This is becoming increasingly apparent to the public everyday. Perhaps the recent emphasis on the concept of similarity will lead us to better understand this fuzzy universe we inhabit. Then again, perhaps the current fad will slow as people take
a hard look at what this approach really does for their work and what practical results emerge from it. The Beilstein Institute has done us all a service by organizing this workshop and it will be interesting to review this area in some
five years time to see if the promise of similarity in chemistry has been fulfilled.
1. For further information concerning the 10th. ICCCRE, please contact the Conference Chair, Prof. Y. Wolman,
Department of Organic Chemistry, Hebrew University of Jerusalem, Jerusalem, 91904, Israel. Telephone: 972-2-585-282. FAX 660-346; 666-804. Internet: CHESKEA@HUJIVMS.
2.Papers presented at the 10th. ICCCRE will be published in the March, 1993 issue of
the Journal of Chemical Information and Computer Sciences.
3. Proc. Beilstein Workshop on the Estimation of Physical Data for Organic Compounds.
May, 1988. Jochum, C.; Hicks, M. G.; Sunkel, J. Springer-Verlag, NY, 1988.
4. J. Chem. Inf. Comp. Sci, 1990, 30, 351-520.
5. Concepts and Applications of Molecular Similarity. Johnson, M. A.; Maggiora, G. M. Wiley, NY, 1990.
6. Rouvray, D. H. The Definition and Role of Similarity Concepts in the Chemical and Physical Sciences. J. Chem. Inf. Comput. Sci, 1992, 32, aaa-bbb.
7.Carbo, R.; Calabuig, B. Quantum Similarity Measures, Molecular Cloud Description
and Structure-Properties Relationships. J. Chem. Inf Comput. Sci, 1992, 32, aaa-bbb.
8. Mezey, P. G. Shape-Similarity Measures for Molecular Bodies: A 3D Topological Approach to QShAR.
J.Chem. Inf. Comput. Sci, 1992, 32, aaa-bbb.
9. Lawson, A. J. Organic Reaction Similarity in Information Processing. J. Chem. Inf Comput. Sci, 1992, 32, aaa-bbb.
10. Ponec, R.; Strnad, M. Similarity Ideas in the Theory of Pericyclic Reactivity. J. Chem. Inf Comput. Sci, 1992, 32, aaa-bbb.
11. Fontain, E. The Application of Genetic Algorithms in the Field of Constitutional Similarity. J. Chem. I~ Comput. Sci, 1992, 32, aaa-bbb.
12. Sello, G. Reaction Prediction: The Suggestions of the "Beppe" Program. J. Chem. Inf Comput. Sci, 1992, 32, aaa-bbb.
13. Gifford, E.; Johnson, M.; Tsai, Chun-che; Kaiser, D. A Visualizable Molecular Similarity Space for Modeling the Relative Metabolic Occurrence of N-Oxidation and N-Demethylation. J. Chem. In~ Comput. Sci, 1992, 32, zaaa-bbb.
14. Gasteiger, J.; Ihlenfeldt, W.-D.; Fick, R.; Rose, J. R. Similarity Concepts for the
Planning of Organic Reactions and Syntheses. J. Chem. Inf. Comput. Sci, 1992,32, aaa-bbb.
15. Hanessian, S.; Botta, M.; Larouche, B.; Boyaroglu, A. Computer-Assisted Percept I..
of Similarity Using the Chiron Program - A Powerful Tool for the Analysis and Prediction of Biogenetic Patterns. J. Chem. Inf. Comput. Sci, 1992, 32, aaa-bbb.
16. Artymiuk, P. J.; Grindley, H. M.; Rice, D. W.; Ujah, E. C.; Willett, P. Similarity Searching in Databases of Three-Dimensional Molecules and Macromolecules. J. Chem. Inf. Comput. Sci., 1992, 32, aaa-bbb.
17. Fisanick, W.; Cross, K. P.; Lillie, D. H.; Lipkus, A. H., Rusinko III, A. A Comparison of Similarity Searching on 2D, 3D and Molecular Property for CAS Registry Substances. J. Chem. Inf. Comput. Sci, 1992, 32, aaa-bbb.
18. Bayada, D.; Simpson, R. W.; Johnson, A. P.; Laurenpo, C. An Algorithm for the Multiple Common Subgraph Problem. J. Chem. Inf. Comput. Sci, 1992, 32, aaa-bbb.
19. Takahashi, Y.; Sukekawa, M.; Sasaki, S.-I. Automatic Identification of Molecular Similarity Using Reduced Graph Representation of Chemical Structure. J. Chem. Inf. Comput. Sci., 1992, 32, aaa-bbb.
20. Marsili, M.; Sailer, H. ANALOGS: A Computer Program for the Design of Multivariate Sets of Analogs. J. Chem. Inf. Comput. Sci, 1992, 32, aaa-bbb.
21. Perry, N. C.; van Geerestein, V. J. 3D Database Searching on the Basis of Molecular Similarity Using the SPERM Program. J. Chem. Inf. Comput. Sci, 1992, 32, aaa-bbb.
22. Barnard, J. M.; Downs, G. M. Clustering of Chemical Structures on the Basis of 2-D Similarity Measures. J. Chem. Inf. Comput. Sci, 1992, 32, aaa-bbb.
23. Judson, P. N. Relating Structural Similarity Searching to Biological Activity. J. Chem. Inf. Comput. Sci, 1992, 32 aaa-bbb.
24. Allan, N. L; Cooper, D. L. A Momentum Space Approach to Molecular Similarity. J. Chem. Inf. Comput. Sci, 1992, 32, aaa-bbb.
25. Randic, M. Similarity Based Upon Extended Basis Descriptors. J. Chem. Inf. Comput. Sci, 1992, 32, aaa-bbb.
26. JCICS issue on Markush Searching.
27. Maggiora, G. M.; Elrod, D. W.; Trenary. Computational Neural Nets as Model-Free
Mapping Devices. J. Chem. Inf. Comput. Sci, 1992, 32, aaa-bbb.