A stepwise approach for determining the model applicability domain is proposed. Four stages are applied to account for the diversity and complexity of the current SAR/QSAR models, reflecting their mechanistic rationality (including metabolic activation of chemicals) and transparency. General parametric requirements are imposed in the first stage, specifying in the domain only those chemicals that fall in the range of variation of the physicochemical properties of the chemicals in the training set. The second stage defines the structural similarity between chemicals that are correctly predicted by the model. The structural neighborhood of atom-centered fragments is used to determine this similarity. The third stage in defining the domain is based on a mechanistic understanding of the modeled phenomenon. Here, the model domain combines the reliability of specific reactive groups hypothesized to cause the effect and the domain of explanatory variables determining the parametric requirements in order for functional groups to elicit their reactivity. Finally, the reliability of simulated metabolism (metabolites, pathways, and maps) is taken into account in assessing the reliability of predictions, if metabolic activation of chemicals is a part of the (Q)SAR model. Some of the stages of the proposed approach for defining the model domain can be eliminated depending on the availability and quality of the experimental data used to derive the model, the specificity of (Q)SARs, and the goals of their ultimate application. The performance of the proposed definition of the model domain is tested using several examples of (Q)SARs that have been externally validated, including models for predicting acute toxicity, skin sensitization, and biodegradation. The results clearly showed that credibility in predictions of QSAR models for chemicals belonging to their domain is much higher than for chemicals outside this domain.

The base-line modeling concept presented in this work is based on the assumption of a maximum bioconcentration factor (BCF) with mitigating factors that reduce the BCF. The maximum bioconcentration potential was described by the multi-compartment partitioning model for passive diffusion. The significance of different mitigating factors associated either with interactions with an organism or bioavailability were investigated. The most important mitigating factor was found to be metabolism. Accordingly, a simulator for fish liver was used in the model, which has been trained to reproduce fish metabolism based on related mammalian metabolic pathways. Other significant mitigating factors, depending on the chemical structure, e.g. molecular size and ionization were also taken into account in the model. The results (r(2)=0.84) obtained for a training set of 511 chemicals demonstrate the usefulness of the BCF base line concept. The predictability of the model was evaluated on the basis of 176 chemicals not used in the model building. The correctness of predictions (abs(logBSF(Obs)-logBCF(Calc))=0.75)) for 59 chemicals included within the model applicability domain was 80%.

The bioconcentration factor (BCF) is a parameter that describes the ability of chemicals to concentrate in aquatic organisms. Traditionally, it is modeled by the log–log quantitative structure -activity relationship (QSAR) between the BCF and the octanol- water partition coefficient (Kow). A significant scatter in the parabolic log(BCF)/log(Kow) curve has been observed for narcotics with log(Kow) greater than 5.5. This study shows that the scatter in the log(BCF)/log(Kow) relationship for highly hydrophobic chemicals can be explained by the molecular size. The significance of the maximal cross-sectional diameter on bioconcentration was compared with the traditionally accepted effective diameter. A threshold value of about 1.5 nm for this parameter has been found to discriminate chemicals with log(BCF) > 3.3 from those with log(BCF) < 3.3. This critical value for the maximum diameter is comparable with the architecture of the cell membrane. This threshold is half thickness of leaflet constituting the lipid bilayer. The existence of a size threshold governing bioconcentration is an indication of a possible switch in the uptake mechanism from passive diffusion to facilitated diffusion or active transport. The value of the transition point can be used as an additional parameter to hydrophobicity for predicting BCF variation. The effect of molecular size on bioconcentration has been studied by accounting for conformational flexibility of molecules.

The bioaccumulation potential of chemicals is used to indicate when chemicals are likely to contaminate fish, birds and other wildlife, and humans. Together with knowledge of the persistence of chemicals, the bioaccumulation potential is useful in setting priorities for hazard identification as well as environmental monitoring. Because the measurement of the bioaccumulation potential is costly, developing reliable estimates of this important indicator directly from chemical structure has long been a goal of Quantitative Structure Activity Relationship (QSAR) practitioners. Many previous models for predicting bioconcentration factors (BCF) for organic chemicals have been based on linear and bilinear relationships between log(BCF) and octanol-water partition coefficient (log(K ow )) , some of which also included other structural parameters such as structural correction factors or molecular connectivity indices, Fujitas characters, etc. Most of these BCF models have been derived for predicting passive diffusion of chemicals with log. octanol-water partition coefficients log(K ow ) < 7. Most previous models showed large discrepancy for large number of chemicals (predominantly highly lipophilic) found in humans and fish. The effect of steric molecular attributes on predicting BCF was studied using 694 chemicals with available experimental BCF and K ow values. It was found that maximum cross sectional diameters and conformational flexibility of chemicals affect significantly bioconcentration and could be used to explain identification of certain highly hydrophobic chemicals in humans and fish.

Traditional attempts to model genotoxicity data have been limited to congeneric data sets, primarily because the mechanism of action was ignored, and frequently, the chemicals required metabolism to the active species. In this exercise, the COmmon REactivity PAtterns (COREPA) approach was used to delineate the structural requirements for eliciting mutagenicity in terms of ranges of descriptors associated with three-dimensional molecular structures. The database used to build the mutagenicity model includes 1196 structurally diverse chemicals tested in the Ames assay by the National Toxicology Program. This manuscript describes the development of the TA100 model that predicts the results of mutagenicity testing using only the Ames TA100 strain. The TA100 model was developed using 148 chemicals that tested positive in TA100 strain without rat liver enzymes (S-9) and 188 chemicals that tested positive in TA100 strain with rat liver enzymes. A decision tree was developed by first comparing the reactivity profile of chemicals that were positive in TA100 without rat liver enzymes to the reactivity profile of the remaining 1048 chemicals. This approach correctly identified 82% of the primary acting mutagens and 94% of the nonmutagens in the training set. The 188 chemicals in the training set that are positive only in the presence of metabolic activation would pass through the decision tree as negative. The next step was to identify the chemicals that are positive only in the presence of metabolic activation. To accomplish this, a series of hierarchically ordered metabolic transformations were used to develop an S-9 metabolism simulator that was applied to each of the 1048 chemicals. The potential metabolites were then screened through the decision tree to identify reactive mutagens. This model correctly identified 77% of the metabolically activated chemicals in a training set. A computer system that applies the COREPA models and predicts mutagenicity of chemicals, including their metabolic activation, was developed. Each prediction is accompanied by a probabilistic estimate of the chemical being in the structural domain covered by the training set.