MP11 Advances in Bioaccumulation Assessment
Monday, 14 November 2005: 8:00 AM - 6:30 PM in Exhibit Hall

(DIM-1117-815796) Applicability Domain of Base Line BCF Model and Quality of Predictions. Screening of Chemicals on DSL.

Mekenyan, O¹, Dimitrov, S¹, Dimitrova, N¹, Stoyanova, G¹, MacDonald, D², Morin, D², ¹ Laboratory of Mathematical Chemistry, University “Prof. As. Zlatarov”, Bourgas, Bulgaria² Existing Substances Branch, Environment Canada, Place Vincent Massey, 351 St. Joseph Boulevard, Gatineau, Quebec, Canada

ABSTRACT- According to base-line concept, the bioaccumulation is defined as a combined effect of passive diffusion (maximum bioaccumulation described by multicompartment partitioning model) and set of mitigating factors associated with bioavailability (size, ionization) and the effect of organism (metabolism). The maximum diameter of molecules is used as a size parameter and the model accounts for their conformational flexibility. A simulator of fish liver metabolism is used to model the metabolic transformations of chemicals. The metabolic pathways and transformation products are predicted explicitly. The thresholds for molecular size, weight of different ionizing groups and metabolic transformation probabilities are parameterized by non-linear least square method to reproduce BCF values and metabolism of 515 chemicals. The performance of the base-line BCF model to reproduce BCF data (R² = 0.84) and metabolism (R² = 0.80) for the training sets was an indication for its usefulness for screening purposes. To ascertain whether the model is applicable or not to a query chemical submitted for screening we have defined its applicability domain. The structural domain of the present BCF model is defined by extracting atom centered fragments from correctly predicted chemicals in training sets. The software implementation of the model provides screening for all chemicals specifying those which are in the model domain. For 59 MITI chemicals identified to belong to the model applicability domain the predictability was 80%. Presently, the model is used to screen the chemicals on DSL. The applicability domain of the current model covers 12% of the structural diversity of DSL inventory. Work on expending model applicability domain of the present model is in progress.

Key words: BCF, QSAR, model applicability domain, base-line concept