Search

The lipophilicity of polyphenols inherent in food, beverages, and medicinal plants was modelled by using 3D descriptors derived from optimized 3D molecular structures in combination with 2D descriptors. The training sets were generated by manual selection or by cluster formation, and statistically robust predictive models were obtained in both cases. The most relevant structural features for the lipophilicity of polyphenols are depicted by the statistically most significant variables: the number of donor atoms for the H bonds is unfavorable for lipophilicity, and the enhanced number of ring secondary C atom (sp3) also decreases lipophilicity, while the increased atomic polarizability implies higher lipophilicity of polyphenols. The study also revealed the importance of a three-dimensional distribution of atomic electronegati-vity for the lipophilicity of molecules. [ABSTRACT FROM AUTHOR]