Use of quantitative structural analysis to predict fish bioconcentration factors for pesticides.

The focus of this research was to develop a model based solely on molecular descriptors capable of predicting fish bioconcentration factors (BCF). A fish BCF database was developed from high-quality, regulatory agency reviewed studies for pesticides based on the same laboratory protocol and the same fish species, Lepomis macrochirus. A commercially available software program was used to create a quantitative structure-activity relationship (QSAR) from 93 BCF studies based on unique molecules. An additional 16 molecules were used to test the accuracy of QSAR model predictions for a variety of pesticide classes. Regression of the measured versus predicted log BCF values yielded a regression coefficient of 0.88 for the validation data set. On the basis of the results from this research, the ability to predict BCF by a QSAR regression model is improved using a fully structurally derived model based solely on structural data such as the number of atoms for a given group (e.g., -CH3) or the local topology of each atom as derived from electron counts. Such descriptors provide insightful information on a molecule's potential BCF behavior in aquatic systems.