The toxicity of 32 shale oil components to a mixed marine bacterial culture was assessed utilizing an acute static bioassay protocol. Structure-activity relationships were examined between toxicity and 39 physicochemical properties and molecular descriptors, many of which have not been previously utilized in quantitative structure-activity relationship (QSAR) research. The toxicity could be predicted with reasonable accuracy using simple linear regression equations. However, multiparametric linear regression (MLR) equations yielded improved correlation coefficients and predictive capabilities. Such equations allow the inclusion of two or more characteristics in one relationship. The best MLR equations combined carbon number and interatomic distance between the most positive and negative charges, solubility, the approximate sigma electron density, and a variety of other parameters. These multiparametric QSARs with high predictive capabilities require computer-generated data which may not be readily available. Other MLR equations combine parameters such as the octanol water partition coefficient and the specific molecular connectivity index. These latter equations have lower correlation coefficients and yet are probably of more practical use as the data required are readily available or can be calculated. The presented multiparametric QSARs are applicable within the log Kow range of 0.6 to 5.5.
ASJC Scopus subject areas
- Public Health, Environmental and Occupational Health
- Health, Toxicology and Mutagenesis