A QSPR Study of Sweetness Potency Using CODESSA Program
Alan R. Katritzky, Ruslan Petrukhin, Subbu Perumal, Mati Karelson, Indra Prakash and Nitin Desai
A total of 396
natural and artificial comprehensively referenced sweeteners were classified by
their structures into nine sets. The sweetness potencies were correlated with
quantum chemical and other molecular descriptors using the heuristic and the
best multi-linear regression methods of the CODESSA software package. QSPR
models (two-parameter unless otherwise indicated) emerged for each subclass of
sweeteners with R2 values of (i) 0.835 for 47 aldoximes, (ii) 0.959
for 8 acesulfamates, (iii) 0.927 for 9 sulfamates, (iv) 0.941 for 10
-arylsulfonylalkanoic
acids, (v) 0.715 for 27 guanidines (0.802 in a three-parameter correlation),
(vi) 0.769 for 30 ureas/thioureas (0.888 in a three-parameter correlation),
(vii) 0.905 for 20 natural sweeteners, (viii) 0.957 for 7 miscellaneous
sweeteners (one- parameter correlation), and (ix) 0.688 for 87 peptides
(five-parameter correlation). A significant global five-parameter QSPR
theoretical model with R2 of 0.686 for the entire set of sweeteners
is presented and discussed with reference to the possible existence of single or
multiple sweetness receptors.