Physicochemical properties/descriptors governing the solubility and partitioning of chemicals in water-solvent-gas systems. Part 1. Partitioning between octanol and air

Prediction of biomagnification factors for some organochlorine compounds using linear free energy relationship parameters and artificial neural networks

Multiple linear regression and artificial neural networks (ANNs) as feature mapping techniques were used for the prediction of the biomagnification factor (BMF) of some organochlorine pollutants. As independent variables, or compound descriptors, the Abraham descriptors often employed in linear free energy relationships were used. Much better results were obtained from the nonlinear ANN model than from multiple linear regression. The average absolute error, average relative error and root mean square error in the calculation of log (BMF) by the ANN model were 0.055, 0.051 and 0.097 for the training set and 0.11, 0.086 and 0.175 for the internal validation set, respectively. The degree of importance of each descriptor was evaluated by carrying out a sensitivity analysis approach for the nonlinear model. The results obtained reveal that the order of importance is the pollutant volume, the pollutant dipolarity/polarizability and the pollutant excess molar refraction. In order to examine the credibility of the obtained ANN model the leave-many-out cross-validation test was applied which gave Q(2)= 0.827 and SPRESS = 0.15. Also the Y-scrambling procedure was applied to the ANN model in order to examine the effect of chance correlations. The results obtained reveal that it is possible to predict the BMFs of organochlorine pollutants using a nonlinear ANN model with Abraham descriptors as inputs.

Physicochemical properties/descriptors governing the solubility and partitioning of chemicals in water-solvent-gas systems. Part 2. Solubility in 1-octanol

On the basis of octanol solubility data (log S(o)) for 218 structurally diverse solid chemicals it was shown that the exclusive consideration of melting points did not provide satisfactory results in the quantitative prediction of this parameter (s = 0.92). The application of HYBOT physicochemical descriptors separately (s = 0.94) and together with melting points (s = 0.70) in the framework of a common regression model also was not successful, although contributions of volume-related and H-bond terms to solubility in octanol were identified. It was proposed that the main reason for such behaviour was the different crystal lattice interaction of different classes of chemicals. Successful calculations of the solubility in octanol of chemicals of interest were performed on the basis of the experimental solubility of structurally/physicochemically/numerically similar nearest neighbours with consideration of their difference in physicochemical parameters (molecular polarisability, H-bond acceptor and donor factors (s = 0.66)) and of these descriptors together with melting point differences (s = 0.38). Good results were obtained for all compounds having nearest neighbours with sufficient similarity, expressed by Tanimoto indexes, and by distances in the scaled 3D descriptor space. Obviously the success of this approach depends on the size of the database.