David W. Elrod

Applications for neural networks in chemistry. 2. A general connectivity representation for the prediction of regiochemistry

Abstract A general method for the prediction of organic reactions by a backpropagation neural network is described. Neural networks trained using modified Dugundji-Ugi BE-matrix representations gave excellent predictions of the regiochemistry for three different types of reactions: Markovnikov addition to alkenes, Diels-Alder and retro-Diels-Alder reactions, and Saytzeff elimination. The networks were able to extract reactivity information from examples of the reactions to develop an internal representation of the reactions without explicitly incorporating rules into the network. Since the neural network was better at interpolating than extrapolating, it is important that the training set span the set of possible reactions. The method of representation used is sufficiently general to handle most classes of organic reactions.

Predicting Chemical Reactions with a Neural Network

This project is an attempt to use neural networks to extract relevant features from data about the reactivity and product distribution in a chemical reaction in order to predict the percentages of products formed in the reaction. Electrophilic aromatic substitution reactions were chosen because they have been well studied and a fairly large amount of data is available about the percent of each of three resultant isomers formed when various chemical substituents were present. Different representation schemes and network architectures were used, and results achieved are comparable to results obtained from existing programs and from the intuition of chemists.

Poster Session: Computer Assisted Synthesis Design Using CHIRON and REACCS

Aranciamycinone, an anthracycline antitumour antibiotic, has three adjacent chiral centres on its A ring. Most of the published syntheses for anthracyclines have resulted in racemic products because of the difficulties inherent in asymmetric synthesis. CHIRON, developed by Steve Hanessian at the University of Montreal, uses a pattern recognition algorithm to relate the stereochemistry of the target to a library of chiral starting materials. CHIRON allows the precursors to be saved in a format compatible with MDL’s REACCS program. REACCS was used to work out a possible pathway for the synthesis of optically active Aranciamycinone starting from chiral precursors suggested by CHIRON.