Predictive modeling of the total deactivation rate constant of singlet oxygen by heterocyclic compounds.

We constructed a predictive model of the total deactivation rate constant (k(t)) of singlet oxygen by heterocyclic compounds that are widespread in biological systems and participate in highly relevant biologic functions related with photochemical processes, by means of quantitative structure-property relationships (QSPR). The study of the reactivity of singlet oxygen with biomolecules provides their antioxidant capability, and the determination of the rate constants allows evaluation of the efficiencies of these processes. Our optimal linear model based on 41 molecular structures, which have not been used previously in a QSPR study, consists of six variables, selected from more than thousand geometrical, topological, quantum-mechanical and electronic types of molecular descriptors. Our recently developed strategy to determine the optimal number of descriptors in model is successfully applied. As a practical application of our QSPR model we estimated the unknown k(t) of several heterocyclic compounds that are of particular interest for further experimental studies in our research group.