Methods of Computer Aided Synthesis Design


Edward (Ned) S. Blurock
(Home Page)
E-Mail:
Edward.Blurock@forbrf.lth.se
Snail-Mail

Combustion Physics
Lund University
P.O. Box 118
SE-221 00 LUND
Sweden
Phone: +46 46 222 1402
FAX: +46 46 0885

This is all the course materials and slides for the summer semester 95 lecture on Methods of Computer Aided Synthesis Design. One of the main emphases of this course will be to introduce the use of symbolic (i.e. non-numeric) methods in computers and chemistry. This course gives an overview of the Computer Aided Organic Synthesis (CAOS) systems that are presently available. The first part of the course will be an introcution to the significant systems. The second part of the course uses the framework of artificial intellegence to introduce some of the methods used in these systems.

Lecture 1: Introduction

The purpose of this lecture is not only for the lecturer to introduce himself (through an outline of his work in this area), but also to acquire a feeling for what approach should be taken in introducing the material in this course.
Handouts:
NONE
Overheads:
Course:
What the course covers, i.e. Symbolic Methods in Chemistry
Background:
An Outline of Blurock's Research in Symbolic Modeling in Chemistry. The emphasis is on the three software systems in CAOS, machine learning and combustion modeling.

Lecture 2: Introduction to REACCS

Handouts:
Molecular Design Ltd.: Reaction Similarity Searching: A New Approach to Reaction Retrieval
A slide presentation of the similarity function in REACCS
Molecular Design Ltd.: Application Notes - Multistep Synthesis Planning
A complete example in use REACCS for multistep synthetic analysis
Overheads:
REACCS Application:
A brief collection of slides accompanying the application notes


Lecture 3: CAOS Systems Overview and Demonstration of Network

Handouts:
Computer Tools for Reaction Retrieval and Synthesis Planning in Organic Chemistry: A Brief Review of their History, Methods and Programs
Martin A. Ott and Jan H. Noordik (CAOS/CAMM Center), Recuil des Travaux des Pay-Bas, 111, 239-2476 (1992)
Overheads:
Review of CAOS Systems:

Network Demo:
Intro To Internet:
Summary of Chemistry Sites:
Organic Chemistry Resources:
Searching The Internet:
Libraries in Austria BIBOS:
CAOS/CAMM Center:
Chemist's Art Gallery:
WWW in the World:
REACCS telnet to Vienna:


Lecture 4: LHASA

A more in depth view of the LHASA system of Corey. LHASA is written as a classical expert system. The chemical knowledge is represented as programmed transforms. The system is examined from the user's standpoint and then, two important aspects, strategies and transform ratings are discussed.
Handouts:






Overheads:
User Standpoint:
A general introduction of LHASA from the user's standpoint
LHASA Strategies:
Strategies in LHASA are reviewed and the starting material strategy is discussed in detail
LHASA Ratings:
How the transforms are initially ordered (the concept of ratings)


Lecture 5: SYNGEN

An overview of the SYNGEN system of Hendrikson. In contrast to LHASA, the chemical knowledge in this system is represented within a logical framework: Atoms are generalized to a few abstracted types and reactions are viewed as a small number of electrophilic and nucleophilic half-reactions. The synthetic strategy pursued is constructive synthesis.
Handouts:
A Logic-Based Program for Synthesis Design
James B. Hendrikson, David L. Grier, A. Glen Toczko; J. Amer. Chem. Society, vol 107, 5228-5238 (1985)
Overheads:
SYNGEN system:
A general introduction of LHASA from the user's standpoint


Lecture 6: EROS

A general overview of the EROS/WODCA system of the Gasteiger group. First the molecular representation with the associated use of semi-empirical (graphical) methods of physical property and reactivity prediction are introduced and then an overview of the current system with its wide selection of methods available for computer-aided synthesis design is presented.
Handouts:
A Collection of Computer Methods for Synthesis Design and Reaction Prediction
J. Gasteiger, W.D. Ihlenfeldt, P. Rose; Recl. Trav. Chim. Pays-Bas, 111, 270-290 (1992)
Acquisition and Representation of Knowledge for Expert Systems in Organic Chemistry
J. Gasteiger, M. G. Hutchings, P. Low, H. Saller; ACS Symposium Series 306, Artificial Intelligence Applications in Chemistry
Overheads:
EROS/WODCA system:
A general introduction of the system of Gasteiger containing a wide range of methods for the CADS
Reaction Prediction :
Physical chemistry and reaction prediction within EROS

Lecture 7: Molecule Representations

Two problems of representation of the molecule within the computer are presented. First, the 2-D, i.e graphical, representation is explained with examples. Then, the problem of labeling the atoms in a canonical way is explained and typical algorithms given. The use of canonical labeling is used as an introduction to the problem of determining whether two molecules are equal or not.
Handouts:
Artificial Intelligence in Organic Chemistry: 2-D Representation:
A short introduction to two graphical representations of molecules
Elementary Problems of Presentation of Structural Codes in Canonical Form
Z. Hippe; Artificial Intelligence in Chemistry: Structure Elucidation and Simulation of Organic Reactions; Chapter 9.1
Overheads:
Canonical Ordering of Atoms in Molecules:

Lecture 8: Graph Theory: Subsubstructure Search

The problem of substructure search is transposed to the problem of subgraph isomorphism in graph theory. Simple algorithms for subgraph search are explained first in a combinatorial way and then within the framework of generalized search within artificial intelligence (AI). The opportunity is used to introduce basic terminology of AI.
Handouts:
Artificial Intelligence in Chemistry: Recognizing Chemical Substructures:

Overheads:
From paper

Lecture 9: Heuristics, Reaction Center and Maximal Common Subgraph

The problem of molecular representation is brought up once again in the sense of representing more global information about the atomic environment within the molecule. The information will be used for establishing heuristics for the more complex problem of finding the reaction center of a molecule (Maximum Common Subgraph).
Handouts:
None

Overheads:
Notes on Reaction Center, Graph Difference, Maximal Common Subgraph and Heuristics in Generalized Search



Lecture 10: Databank Analysis: Machine Learning of Chemical Concepts

The basic problem of representing the molecule for qualitative and quantitative structure-property relationships (QSPR) is introduced with an example (by the author) of using several machine learning techniques to determine Octane Number from structural features. A short introduction to using such techniques for reaction characterisation is also introduced.
Handouts:
Automatic Learning of Chemical Concepts: Research Octane Number and Molecular Substructures
Edward S. Blurock; Computers and Chemistry 19, 91-99 (1995)
Computer-Aided Synthesis Design at RISC-Linz: Automatic Extraction and Use of Reaction Classes
Edward S. Blurock; J. Chem. Inf. and Comp. Sci. 30, 505-510 (1990)
Edward.Blurock@forbrf.lth.se