Development of structure-taste relationships for sweet and non-sweet heterosulfamates

A data set of 101 hetero- (both cyclic and open chain) sulfamate sodium salts, whose taste data are known, have been assembled and divided into sweet (S) (20 compounds) and non-sweet (N) (81 compounds) categories. The data set is made up of 56 compounds reported earlier, 32 synthesised in this work and another 13 reported since the earlier publications. Using the parameters x, y and z (measured for the RNH portion of RNHSO3Na using CPK models) and first order molecular connectivity, (1)chi(nu) it has been possible to achieve a correct classification rate of approximately 65% using linear discriminant analysis (LDA): a compound is N if 3.285+0.439x+0.662y+ 0.236z-1.27 (1)chi(nu) 0 otherwise it would be S. Using quadratic discriminant analysis (QDA) the classification rate increased to approximately 80%. Finally a Tree-based analysis gave an 86% classification rate but performed poorly in classifying correctly the S group of compounds.