Study on octanol-water partition coefficients of benzene halides by molecular vertex distance index

Abstract

A new method of molecular structural characterization (MSC) called the molecular vertex distance index (MVDI) is constructed and used to describe the structures of benzene halides. Two quantitative structureproperty relationship (QSPR) models of octanol-water partition coefficients (lgKow) are obtained through multiple linear regression (MLR) and partial least squares regression (PLSR). The estimation stability and generalization ability of the models are analyzed by both internal and external validations. The results show that the models constructed in this work can provide satisfactory estimation stability and favorable predictive ability.The hydrophobic constant of organic compounds plays very important role in the environmental risk assessment and it is directly related to the migration and distribution of pollutants in the atmosphere, water, sediments, soil, and other environmental media as well as the organisms. The octanol-water partition coefficient (lgKow) is the most common form of the hydrophobic constant of organic compounds. The construction of a quantitative structure-property relationship (QSPR) requires the structural descriptors of related chemical compounds. Much meaningful work on QSPR has been done for a long time [1-3]. We took 38 benzene halides as the samples of this study and their lgKow listed in Table 1 were taken from the literature [4]. As for the QSPR research, our laboratory has developed a molecular structural characterization method called the molecular electronegativity-distance vector (MEDV) [5][6][7][8]. Based on the MEDV, the main aim of this study was to develop another molecular structural characterization method called the molecular vertex distance index (MVDI). A useful QSPR model of lgKow of benzene halides was also established in this work.Principles and methods. The property of an organic compound is directly related to the molecular structure. The octanol-water partition coefficient of organic matter is usually affected by the molecular size, flexibility and polarity, the hydrogen bonding between the molecules and so forth. For the molecular matrix skeleton, each non-hydrogen atom is a molecular vertex. The vertex is the basic unit of a molecule, so the external property of a compound can be reflected from the level of its vertices. Hence, the characteristic value of the vertex should be defined firstly for the correlative index. This definition is not a definite restriction and its purpose is to correlate the defined characteristic value with the practical problems of the research. The characteristic value of the vertex should exhibit the atomic structure and its surrounding environment. Without regard to non-framework hydrogen atoms, these vertex atoms were classified as four atomic types according to the atom number of each atom connecting to through some chemical bond/bonds. If an atom is linked to k