Designing Biodegradable Molecules from the Combined Use of a Backpropagation Neural Network and a Genetic Algorithm

Publisher Summary This chapter elaborates a hybrid system constituted of a backpropagation neural network (BNN) and a genetic algorithm (GA) for designing organic molecules presenting a specific biodegradability. The BNN model is derived from a training set, which consisted of the collective judgments of 22 experts as to the approximate time that might be required for aerobic ultimate degradation in receiving waters (AERUD) of 38 highly diverse molecules. Chemicals are described by means of 14 molecular descriptors. The selected 14/2/1 BNN model correctly classified 45 of 49 chemicals (91.8%) in the testing set. The configuration of the GA is optimized for selecting candidate molecules having low AERUD values. Different constraints are added in the GA to test the limits of the hybrid system for proposing biodegradable molecules presenting specific structural features.

[1]  Jan Chudoba,et al.  Biodegradability of Organic Substances in the Aquatic Environment , 1990 .

[2]  D. E. Goldberg,et al.  Genetic Algorithms in Search, Optimization & Machine Learning , 1989 .

[3]  Sukhdev Khebbal,et al.  Intelligent Hybrid Systems , 1994 .

[4]  J Devillers,et al.  Estimating pesticide field half-lives from a backpropagation neural network. , 1993, SAR and QSAR in environmental research.

[5]  Robert S. Boethling,et al.  Screening-level model for aerobic biodegradability based on a survey of expert knowledge , 1989 .

[6]  M. Chastrette,et al.  Multilayer Neural Networks Applied to Structure-Activity Relationships , 1991 .

[7]  W. Meylan,et al.  Development of a predictive model for biodegradability based on BIODEG, the evaluated biodegradation data base. , 1991, The Science of the total environment.

[8]  J. Devillers,et al.  New Trends in Structure‐Biodegradability Relationships , 1993 .

[9]  J Devillers.  Neural modelling of the biodegradability of benzene derivatives. , 1993, SAR and QSAR in environmental research.

[10]  R. Boethling,et al.  Expert systems survey on biodegradation of xenobiotic chemicals. , 1989, Ecotoxicology and environmental safety.

[11]  Robert S. Boethling,et al.  Predictive model for aerobic biodegradability developed from a file of evaluated biodegradation data , 1992 .

[12]  Robert S. Boethling,et al.  Group contribution method for predicting probability and rate of aerobic biodegradation. , 1994, Environmental science & technology.

[13]  Gerald J. Niemi,et al.  Structural features associated with degradable and persistent chemicals , 1987 .

[14]  Russell C. Eberhart,et al.  Neural network PC tools: a practical guide , 1990 .