Interdisziplinare Systemforschung Modelling Analyse ...3A978-3-0348-5788-8%2F1.pdf · lem-solving activity of the human organic synthetic chemist.4,5 In all of these computer applications

Interdisciplinary Systems Research AnalysIs ~ Modelling ~ SimulatIOn

The system science has been developed from several scientific fields: control and communication theory, model theory and computer science. Nowadays it fulfills the requirements which Norbert Wiener formulated originally for cybernetics; and were not feasible at his time, because of insufficient development of computer science in the past. Research and practical application of system science involve works of specialists of system science as well as of those from various fields of application. Up to now, the efficiency of this co-operation has been proved in many theoretical and practical works. The series 'Interdisciplinary Systems Research' is intended to be a source of information for university students and scientists involved in theoretical and applied systems research. The reader shall be informed about the most advanced state of the art in research, application, lecturing and meta theoretical criticism in this area. It is also intended to enlarge this area by including diverse mathematical modeling procedures developed in many decades for the description and optimization of systems. In contrast to the former tradition, which restricted the theoretical control and computer science to mathematicians, physicists and engineers, the present series emphasizes the interdisciplinarity which system science has reached until now, and which tends to expand. City and regional planners, psychologists, physiologists, economists, ecologists, food scientists, sociologists. political scientists, lawyers, pedagogues, philologists, managers, diplomats, military scientists and other specialists are increasingly confronted or even charged with problems of system science. The ISR series will contain research reports - including PhD-theses -lecture notes, readers for lectures and proceedings of scientific symposia. The use of less expensive printing methods is provided to assure that the authors' results may be offered for discussion in the shortest time to a broad, interested community. In order to assure the reproducibility of the published results the coding lists of the used programs should be included in reports about computer simulation. The international character of this series is intended to be accomplished by including reports in German, English and French. both from universities and research centers in the whole world. To assure this goal, the editors' board will be composed of representatives of the different countries and areas of interest.

Editor/ Herausgeber: Prof. Salomon Klaczko-Ryndziun, Frankfurt a. M.

Co-Editors / Mitherausgeber: Prof. Ranan Banerji, Temple University, Philadelphia Prof Jerome A. Feldman, University of Rochester, Rochester Prof Mohamed Abdelrahman Mansour, ETH, Ziirich Prof. Ernst Billeter, Universitat Fribourg, Fribourg Prof Christof Burckhardt, EPF, Lausanne Prof Ivar Ugi, Technische Universitiit Miinchen Prof King-Sun Fu, Purdue University, West Lafayette

Interdisziplinare Systemforschung Analyse ~ Formallslerung ~ SimulatIOn

Die Systemwissenschaft hat sich aus der Verbindung mehrerer Wissenschaftszweige entwickelt: der Regelungs- und Steuerungstheorie, der Kommunikationswissenschaft, der Modelltheorie und der Informatik. Sie erfiillt heute das Programm, das Norbert Wiener mit seiner Definition von Kybernetik urspriinglich vorgelegt hat und dessen Durchfiihrung zu seiner Zeit durch die noch ungeniigend entwickelte Computerwissenschaft stark eingeschrankt war. Die Forschung und die praktische Anwendung der Systemwissenschaft bezieht heute sowohl die Fachleute der Systemwissenschaft als auch die Spezialisten der Anwendungsgebiete ein. In vielen Bereichen hat sich diese Zusammenarbeit mittlerweile bewahrt. Die Reihe ,dnterdisziplinare Systemforschung» setzt sich zum Ziel, dem Studenten, dem Theoretiker und dem Praktiker iiber den neuesten Stand aus Lehre und Forschung, aus der Anwendung und der metatheoretischen Kritik dieser Wissenschaft zu berichten. Dieser Rahmen soli noch insofern erweitert werden, als die Reihe in ihren Publikationen die mathematischen MOdellierungsverfahren mit einbezieht, die in verschiedensten Wissenschaften in vielen Jahrzehnten zur Beschreibung und Optimierung von System en erarbeitet wurden. Entgegen der friiheren Tradition, in der die theoretische Regelungs- und Computerwissenschaft auf den Kreis der Mathematiker, Physiker und Ingenieure beschrankt war, liegt die Betonung dieser Reihe auf der Interdisziplinaritat, die die Systemwissenschaft mittlerweile erreicht hat und weiter anstrebt. Stadt- und Regionalplaner, Psychologen, Physiologen, Betriebswirte, Volkswirtschafter, Okologen, Ernahrungswissenschafter, Soziologen, Politologen, Juristen, Padagogen, Manager, Diplomaten, Militarwissenschafter und andere Fachleute sehen sich zunehmend mit Aufgaben der Systemforschung konfrontiert oder sogar beauftragt. Die ISR-Reihe wird Forschungsberichte - einschliesslich Dissertationen -, Vorlesungsskripten, Readers zu Vorlesungen und Tagungsberichte enthalten. Die Verwendung wenig aufwendiger Herstellungsverfahren soli dazu dienen, die Ergebnisse der Autoren in kiirzester Frist einer moglichst breiten, interessierten Offentlichkeit zur Diskussion zu stellen. Um auch die Reproduzierbarkeit der Ergebnisse zu gewahrleisten, werden in Berichten iiber Arbeiten mit dem Computer wenn immer moglich auch die Befehlslisten im Anhang mitgedruckt. Der internationale Charakter der Reihe soli durch die Aufnahme von Arbeiten in Deutsch, Englisch und Franzosisch aus Hochschulen und Forschungszentren aus aller Welt verwirklicht werden. Dafiir soli eine entsprechende Zusammensetzung des Herausgebergremiums sorgen.













ISR23 Interdisciplinary Systems Research Interdisziplinare Systemforschung



Henry W. Davis

Computer Representation of the Stereochemistry of Organic Molecules

With application to the problem of discovery of organic synthesis by computer

Springer Basel AG 1976

Henry W. Davis




Henry W. Davis




CIP-Kurztitelaufnahme der Deutschen Bibliothek

Davis, Henry M. Computer representation of the stereochemistry of organic molecules: with application to the problem of discovery of organic synthesis by computer. — 1 .Aufl. — Basel, Stuttgart: Birk-häuser, 1976.

(Interdisciplinary systems research; 23) ISBN 978-3-7643-0847-6 ISBN 978-3-0348-5788-8 (eBook) DOI 10.1007/978-3-0348-5788-8

All rights reserved. No part of this publication may be reproduced stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without the prior permission of the copyright owner.

© Springer Basel AG 1976 Ursprünglich erschienen bei Birkhäuser Verlag Basel 1976

To Herbert Gelernter

His aid, encouragement and inspiration have made this

work possible.



work possible.



work possible.

iii

PREFACE

The role of the computer in the practice of organic

chemistry has been firmly established over the past decade. Its

uses as a large scale information storage and retrieval device

in chemistry have been too numerous to mention. More recently,

the applicability of computers to the problem of discovering

valid and reasonable synthesis routes for organic molecules

has been demonstrated. This has been both as an adjunct to the

chemist in the on-line interactive mode 1,2,3 and also as a wholly

computer-directed system seeking to simulate the intelligent prob-

lem-solving activity of the human organic synthetic chemist. 4 ,5

In all of these computer applications to organic chemistry, it

has been necessary to devise some computer-compatible represen-

tation of an organic molecule that is both canonical and con-

venient for table look-ups. This is in order that entities

that have been constructed at different times under different

circumstances can be identified and classified, with identical

molecules being recognized as such even if their connection

matrices list the elements of the molecule in different orders.

2

3

4

5

E. J. Corey and W. T. Wipke, Science, 166, 178 (1969).

E. J. Corey, W. T. Wipke, R. D. Cramer III and W. J. Howe, J. Americ. Chern. Soc., 94, 421 (1972) and 431 (1972).

E. J. Corey, R. D. Cramer III and W. J. Howe, ~. Americ. Chern. Soc., 94, 440 (1972).

H. L. Gelernter, N. S. Sridharan and A. J. Hart, Topics in Current Chemistry, Vol. 41 (1973), Springer-Verlag.

I. Ugi and J. Dugundji, Topics in Current Chemistry, Vol. 39 (1973), Springer-Verlag.

iii

PREFACE



















2

3

4

5






iii

PREFACE



















2

3

4

5






iv

The canonical representation problem has been satisfac-

torily managed in many different ways where only the constitu-

tional (i.e., topological) structure of the molecule is re-

quired. 6 '7'S Providing a computer-compatible canonical represen-

tation of the stereochemistry of the molecule, however, has been

a far more difficult problem. The problem is a crucial one for

many of the applications mentioned above. It is of particular

importance in the case of wholly computer-directed non-interactive

synthesis discovery systems because the stereochemistry of the

reactants is often a determining factor in deciding whether or not

a given reaction will proceed as desired. In this situation, there

is not chemist on line to build stick models to settle such ques-

tions. Moreover, for many target compounds of biochemical inter-

est, only a particular stereoisomer exhibits the required proper-

ties. It is important in such cases that synthetic pathways be

discovered that will maximize the yield of the desired stereo-

isomer. Unless a stand-alone synthesis discovery program is able

to represent and manipulate the stereochemistry of an organic

structure as readily as it does the topological structure, its

applications will be severely limited, and indeed, such a program

is not likely to attract the serious attention of most organic

chemists.

6

7

S

H. L. Morgan, l. Chern. Doc., 5, 107 (1965).

W. J. Wiswesser, Comp~t. Automat., 19, 2 (1970).

E. G. Smith, The Wiswesser McGraw-Hill, ~Y., 1968.

Line-formula Chemical Notation,

iv






















chemists.

6

7

S





iv






















chemists.

6

7

S





v

The ideas of this book were developed precisely to meet

the needs of such a stand-alone synthesis discovery program. A

quite simple computer-compatible method of representing mole-

cular stereochemistry is described. The method allows straight-

forward identification of such things as which atoms of a mole-

cule are stereochemically indistinguishable and what is a given

molecule's mirror image. Many examples as well as proofs of the

algorithms are included. The algorithms have been implemented

in the synthesis search program SYNCHEM9 , developed under the

direction of Professor H. L. Gelernter at the State University

of New York at Stony Brook.

Several people have been of substantial help during

preparation of this book. The author wishes to express his

gratefulness to Krishna Agarwal for many enlightening con-

versations. Bill Feld and Bob Bingenheimer supplied excellent

ideas concerning the art and did the art work. Cheryl Conrad

and Carol Chandler were invaluable for their excellent

technical typing.

9 H. L. Gelernter, N. S. Sridharan and A. J. Hart, Topics in Current Chemistry, Vol. 41 (1973), Springer-Verlag.

v


















technical typing.


v


















technical typing.


vi

Introduction for the non-chemically trained reader

The non-chemically trained reader should have little dif

ficulty reading this book once he is aware of a few simple

facts and terms. The author, himself, is untrained in chemistry

and approached the problem discussed here as one of information

representation and manipulation. The relevant information was

provided by the chemists.

Roughly, the problem is to find a "convenient" method for

the computer to keep track of how a complicated organic

molecule's atoms are oriented in three-dimensional space. The

method should allow for easy calculation of useful information-

such as which atoms are "look-alikes." Some configurations of

atoms in a molecule will bend and swivel in all sorts to

directions. Others remain relatively fixed. For example, we

can say little about the direction of bond 1 attaching the CH3

group to the oxygen atom in Figure A. On the other hand, the

carbon atom at node 1 in figure A will tend to have its four

ligends (neighbors) attached so that they lie at the corners

of a tetrahedron. If we interchange two of the atoms, say the

chlorine and bromine, without interchanging the other two, the

second tetrahedral configuration cannot be made to coincide

with the first. The two molecules are different. One says

that they are stereoisomers; their connectivity is the same but

their three-dimensional orientation is different. Molecules

whose connectivity descriptions are identical are said to be

constitutionally equivalent. Nodes 2 and 3 in Figure A are

vi



























vi



























vii

not a source of stereoisomerisms because in each case two or

more ligands attached to the node are identical.

Chemists often draw pictures of the three-dimensionality

of molecules using wedges, solid and dotted lines. A solid line

indicates a bond lying in the plane of the paper. A dotted line

indicates a bond extending beneath the plane of the paper. The

thick part of a wedge indicates which of two bonded atoms is

nearest the viewer--typically it indicates an atom sticking out

of the plane of the paper towards the viewer. For example, the

two stereoisomers of the molecule in Figure A are shown in

Figures Band C. In these figures the carbon and hydrogen atoms

are connected to node 1 by bonds in the plane of the paper. In

Figure B the chlorine atom extends towards the reader and the

bromine atom away from the reader. We have not shown the

tetrahedral configuration of nodes 2 and 3 in Figures B, C

because, as was mentioned earlier, the tetrahedral orientation

at these centers is not a source of stereoisomerism. Notice

that the molecules of Figures B, C are mirror images of each

other. One says that the given molecule is chiral because it

differs from its mirror image. The molecules Band C are said

to be chiral antipodes. The carbon atom at node 1 is said to

be a center of assymetry for the molecule of Figure A (or B).

If one of the hydrogens connected to node 2 were replaced by a

bromine atom, then the carbon at node 2 would become another

center of assymetry and the given molecule would have four

stereoisomers--two pairs of chiral antipodes.

vii



























vii



























viii

The reader can now gather that a large molecule may have

tens, even scores, of stereoisomers and the problem of repre-

senting them uniquely and efficiently in the computer becomes

challenging.

There is another structure--the olefin bond--which contributes

to the total three dimensional set of a molecule. When two

carbon atoms are connected by a double bond as in figure D, the

whole structure tends to lie in a plane. Thus the molecules

of Figures D and E are not the same. Chemists speak of such

structures as being a source of geometric isomerism. An

assymetric center such as node 1 in Figure A is a source of

stereoisomerism and is chiral. The molecules of Figures D and E

are achiral, that is, each is identical to its mirror image.

Geometric isomerism and stereoisomerism together contribute to

the total stereochemistry of a molecule. The molecule of Figure

F has both types of isomerism. It represents one of four

"stereoisomers."* Finally, when three carbon atoms are connected

by two olefin bonds, the four ligends at the two ends tend to

form a tetrahedral configuration. This is shown by the molecule

in Figure G and its stereoisomer in Figure H. More examples

of all of these phenomenon may be found in Figures 5.11 through

5.16 where a number of molecules and all their stereoisomers

are depicted.

With these terms and concepts in mind, the non-chemically

trained reader should be able to follow all of the main ideas

in this book.

*To the author it seems that "geo-stereoisomer" or "3Disomer" would be better here. But such words are not used.

viii




challenging.



















are depicted.



in this book.


viii




challenging.



















are depicted.



in this book.


ix

Br H CI _ 6 __ I /" NOde~2 Bond I

/

1 C-O

H ~ " /H

Nadel H/C

''4

" - Node :3 H

Figure A

~r H I \ H

CI-C--C / I \-O-C-H

H H \ H

Figure B

CI H : H Br ~CI 1 ___ 0 / --C -C

I I / ~H H H H

Figure C

F " /Br

C --/ -C

H " CI

Figure D

ix


/

1 C-O

H ~ " /H

Nadel H/C

''4

" - Node :3 H

Figure A

~r H I \ H

CI-C--C / I \-O-C-H

H H \ H

Figure B

CI H : H Br ~CI 1 ___ 0 / --C -C

I I / ~H H H H

Figure C

F " /Br

C --/ -C

H " CI

Figure D

ix


/

1 C-O

H ~ " /H

Nadel H/C

''4

" - Node :3 H

Figure A

~r H I \ H

CI-C--C / I \-O-C-H

H H \ H

Figure B

CI H : H Br ~CI 1 ___ 0 / --C -C

I I / ~H H H H

Figure C

F " /Br

C --/ -C

H " CI

Figure D

x

H Br

"'C=== C/ F / ""CI

Figure E

Figure F

Figure G

Figure H

x

H Br

"'C=== C/ F / ""CI

Figure E

Figure F

Figure G

Figure H

x

H Br

"'C=== C/ F / ""CI

Figure E

Figure F

Figure G

Figure H

Section

1

2

3

4

5

6

Contents

Introduction • . . •

Brief summary of the paper. Other approaches to the problem. The present approach: summary, comparisons

and limitations.

Constitutional Equivalence

Basic terminology and concepts.

Identifying and numbering the CE classes: algorithm 1 • . • . • . • • . •

The atoms of a molecule may be divided into classes of constitutionally equivalent members. An algorithm is given for identifying these classes and numbering them canonically.

xi

Page

1

22

26

The canonical TSD: Algorithm 2 • • • . • • • 43

An algorithm is given which associates with each molecule a canonical incidence-type matrix. The matrix reflects the constitutional structure of the molecule.

Stereochemical equivalence and the canonical parity vector. • . . • • •••

An algorithmic means is given for associating with each molecule a canonical parity vector. This is a sequence of numbers which reflects the molecule's stereochemistry. It may be used for cataloguing and table look-ups. A number of examples are given.

Identifying and numbering the SE classes

The atoms of a molecule may be divided into classes of stereochemically equivalent members, that is, members which are indistinguishable from one another on the basis of the molecule's constitution and stereochemistry. An algorithmic means is given for identifying these classes and numbering them canonically.

• • • • • • 56

• • • • • . 100

Section

1

2

3

4

5

6

Contents



and limitations.





xi

Page

1

22

26







• • • • • • 56

• • • • • . 100

Section

1

2

3

4

5

6

Contents



and limitations.





xi

Page

1

22

26







• • • • • • 56

• • • • • . 100

Bibliography . • .

Appendix: Current algorithms used in SYNCHEM--and extensions . . . .

A family of algorithms is given anyone of which may be used to implement the ideas presented earlier. Emphasis is placed on algorithms currently being used by the computer synthesis search program called SYNCHEM.

Author Index •

General Index

xii

Page

• 118

• • • 119

129

130




Author Index •

General Index

xii

Page

• 118

• • • 119

129

130




Author Index •

General Index

xii

Page

• 118

• • • 119

129

130

Documents

Interdisziplinare Systemforschung Modelling Analyse ...3A978-3-0348-5788-8%2F1.pdf · lem-solving activity of the human organic synthetic chemist.4,5 In all of these computer applications