Semiempirical Topological Index: A Novel Molecular Descriptor for Quantitative Structure-Retention Relationship Studies
BioChem Press, Berenice da Silva Junkes, Renata Dias de Mello, Castanho Amboni, Vilma Edite, Fonseca Heinzen · 2003
Motivation. An important property that has been extensively studied in QSPR is the chromatographic retention. Based on new considerations about the chromatographic behavior and experimental data, our group has developed a new topological index designed semi–empirical topological index, IET. The main goal of the present paper is to generalize the semi–empirical topological index, verifying the predictive–ability of the chromatographic retention for a diverse set of organic compounds (alkanes, alkenes, esters, ketones, aldehydes, and alcohols) and to obtain a general QSRR model. QSRR may be used as an important complementary tool for the elucidation of the molecular structure or for the prediction of the chromatographic retention. Method. This index is based on the hypothesis that the chromatographic retention is due to the interaction of each atom of the molecule with the stationary phase, and consequently the value of the index is reduced by steric effects from its neighbors. Considering that the complexity involved in the solute–stationary phase interactions cannot be estimated only by theoretical considerations, values were attributed to the atoms of the molecules from the experimental chromatographic retention and theoretical deductions. Results. The simple linear regression between the chromatographic retention and the index proposed, for all 548 organic compounds, is extremely satisfactory (correlation coefficient, r = 1.0000, standard deviation, SD = 7.01, and leave–one–out cross–validation correlation coefficient, r 2 CV = 0.999). The predictive quality of the QSRR was tested for an external prediction set of 182 compounds randomly chosen from 548 compounds (r = 1.0000 and SD = 7.65). Conclusions. Statistical analysis shows that the semi–empirical topological index has excellent predictive power using a single descriptor for a large data set of organic compounds.