QSAR models for predicting octanol/water and organic carbon/water partition coefficients of polychlorinated biphenyls

Abstract

Quantitative structure–property relationship modelling can be a valuable alternative method to replace or reduce experimental testing. In particular, some endpoints such as octanol–water (K_OW) and organic carbon–water (K_OC) partition coefficients of polychlorinated biphenyls (PCBs) are easier to predict and various models have been already developed. In this paper, two different methods, which are multiple linear regression based on the descriptors generated using Dragon software and hologram quantitative structure–activity relationships, were employed to predict suspended particulate matter (SPM) derived log K_OC and generator column, shake flask and slow stirring method derived log K_OW values of 209 PCBs. The predictive ability of the derived models was validated using a test set. The performances of all these models were compared with EPI Suite™ software. The results indicated that the proposed models were robust and satisfactory, and could provide feasible and promising tools for the rapid assessment of the SPM derived log K_OC and generator column, shake flask and slow stirring method derived log K_OW values of PCBs.

Keywords:

• Quantitative structure–property relationship
• polychlorinated biphenyls
• octanol–water partition coefficient
• organic carbon–water partition coefficient
• hologram quantitative structure–activity relationship

Acknowledgements

The authors gratefully acknowledge the collaboration of Dr Jintao Yuan (School of Public Health, Zhengzhou University).