摘要
Using the molecular electronegativity distance vector descriptors derived directly from the molecular topological structures, the aqueous solubilities of polychlorinated biphenyls (PCBs) were predicted. A three-variable regression equation with correlation coefficient of 0.9739 and the root mean square errors of 0.26 was developed. The descriptors included in the equation represent three interactions between three pairs of atomic types, i.e., atom -C= and >C=, -C= and -Cl, and -Cl and -Cl. It has been proved that the aqueous solubilities of 137 PCB congeners can be accurately predicted as long as there are more than 65 calibration compounds.
Using the molecular electronegativity distance vector descriptors derived directly from the molecular topological structures, the aqueous solubilities of polychlorinated biphenyls (PCBs) were predicted. A three-variable regression equation with correlation coefficient of 0.9739 and the root mean square errors of 0.26 was developed. The descriptors included in the equation represent three interactions between three pairs of atomic types, i.e., atom -C= and >C=, -C= and -Cl, and -Cl and -Cl. It has been proved that the aqueous solubilities of 137 PCB congeners can be accurately predicted as long as there are more than 65 calibration compounds.