Correlation of the Aqueous Solubility of Hydrocarbons and Halogenated Hydrocarbons with Molecular Structure

Abstract

The aqueous solubilities of a set of 109 hydrocarbons and 132 halogenated hydrocarbons (total 241) are correlated by a three term equation using descriptors calculated solely from molecular structure, with a correlation coefficient (R) of 0.979 and a standard error (s) of 0.386 log units. This equation allows the estimation of aqueous solubilities of hydrocarbons and halogenated hydrocarbons (including polychlorinated biphenyls). The key descriptor is the molecular volume, modified by topological and electrostatic terms. The use of descriptors calculated only from molecular structure eliminates the need for experimental determination of properties for use in the correlation and allows for the estimation of aqueous solubility for molecules not yet synthesized or isolated.

A Fuzzy ARTMAP Based on Quantitative Structure−Property Relationships (QSPRs) for Predicting Aqueous Solubility of Organic Compounds

“…The present models were also compared to recent multilinear regression QSPRs developed by Katritzky et al 16 and Huibers and Katritzky. 15 For the 287 compounds common with the study of Katritzky et al, 16 the present backpropagation QSPR model performed with average and maximum absolute errors of 0.25 and 2.4 logS units, respectively, relative to the corresponding errors of 0.47 and 3.8 logS units for the Katritzky et al 16 model. The present fuzzy ARTMAP was a significant improvement with average absolute and maximum absolute errors of 0.027 and 0.67 logS units, respectively, for the same 287 compound data set.…”

“…For example, Sutter and Jurs 12 reported an absolute error of 0.222 logS units, based on a 9-3-1 neural network based logS model developed using a less diverse set of 140 compounds (-10.83 e logS e 0.28), with an even lower error of 0.197 logS units when polychlorinated biphenyls (PCB) were removed from the data set. 12 Specific performance comparisons of the present models to previously published neural network-based QSPRs for aqueous solubility at 25°C, 12,13,15,16 based on a sets of compounds common with the present study, are provided in Table 5 and Figure 7. The present models were also compared to recent multilinear regression QSPRs developed by Katritzky et al 16 and Huibers and Katritzky.…”

“…14 A more recent study by Huibers and Katritzky 15 reported a logS correlation for 109 hydrocarbons and 132 halogenated hydrocarbons (-10.41e logSe 0.51) with a set of molecular descriptors that included molecular volume, bonding information content 33 (a topological descriptor encoding both the degree of branching and number of different atom types in the molecule), and atomic charge weighted partial negative surface area 22 (sum of atomic surface areas weighted both by atomic surface charge and partial atomic charge for negatively charged surfaces only); the above model performed with a standard error of 0.386 logS units. In a related study, Katritzky et al 16 proposed a linear correlation model for logS, based on 411 compounds (-6.44 e logS e 1.57) using the following six descriptors: a most negative partial charge, relative negative charged surface area, number of electrons, average bond order of nitrogen, zero-order structural information content, and fractional hydrogen donor surface area; the above model performed with a standard error of 0.573 logS units.…”

The Present Utility and Future Potential for Medicinal Chemistry of QSAR/QSPR with Whole Molecule Descriptors

“…The descriptors from the equation #37 account for the dispersion energy of polar solutes in solution, the electrostatic part of the solutesolvent interaction and hydrogen-bonding interactions in liquids. In subsequent studies, the solubility of liquids and soils was described by a three-parameter equation developed from a set of 96 hydrocarbons and 126 halogenated hydrocarbons (excluding compounds capable of forming hydrogen bonds) ( Table 3.1: #38) [99]. The key descriptor in equation #38 was the molecular volume, employed together with additional topological and constitutional descriptors.…”

Predicting Aqueous Solubility of Chlorinated Hydrocarbons by the MCI Approach

“…The first approach [4][5][6][7][8] is to build model from more easily measured physicochemical properties, such as melting point, boiling point, molar volume, partition coefficient, chromatographic retention time, etc. The other method is based on the information from the molecular of the organic chemicals, which can be further divided into two classes, one is group contributions method [9][10][11][12] and the other is QSPR approach [1,[13][14][15][16][17][18][19][20][21][22][23][24][25][26][27][28][29][30][31][32].…”