Pattern Recognition Applications in Chemistry and Pharmacology. II. Pattern Analysis of Substituted Glycolic Acid Esters (Glycolates).

摘要

The pattern recognition technique of Average Information Gain (AIG) was applied to 16chemical structure descriptors (features) which,on the basis of expert opinion,are required for substituted glycolic acid esters (glycolates) to have some type of physiological activity. The AIG method isolated five of the features with a high discriminatory potential. Three of the features were formulated into a decision rule which was used to classify 511 glycolates according to two levels of safety ratio (LD50/MED50) in mice; a correct classification of 301of the compounds was obtained; 88were misclassified and 122were not assigned. A decision rule incorporating all five of the isolated features gave a correct safety ratio classification for 352of 497 glycolates; 89were misclassified and 58were not assigned. Use of the three-feature decision rule for classification of 511 glycolates according to one of two levels of lethality (LD50) in mice correctly classified 263of the compounds in assigning 372to one of the two classes. These results demonstrated the feasibility of using pattern recognition techniques to enhance the probability of finding a compound with desired level of activity and refined estimates of response. The methodology appears applicable to any chemical/pharmacological structure-activity problem. (Author)