Quantitative Structure-Activity Relationship Study on Fish Toxicity of Substituted Benzenes

Use of artificial neural networks in the QSAR prediction of physicochemical properties and toxicities for REACH legislation

With the introduction of the REACH legislation in the European Union, there is a requirement for property and toxicity data on chemicals produced in or imported into the EU at levels of 1 tonne/year or more. This has meant an increase in the in silico prediction of such data. One of the chief predictive approaches is QSAR (quantitative structure-activity relationships), which is widely used in many fields. A QSAR approach that is increasingly being used is that of artificial neural networks (ANNs), and this chapter discusses its application to the range of physicochemical properties and toxicities required by REACH. ANNs generally outperform the main QSAR approach of multiple linear regression (MLR), although other approaches such as support vector machines sometimes outperform ANNs. Most ANN QSARs reported to date comply with only two of the five OECD Guidelines for the Validation of (Q)SARs.

Quantitative structure-activity relationship to predict acute fish toxicity of organic solvents

REACH regulation requires ecotoxicological data to characterize industrial chemicals. To limit in vivo testing, Quantitative Structure-Activity Relationships (QSARs) are advocated to predict toxicity of a molecule. In this context, the topic of this work was to develop a reliable QSAR explaining the experimental acute toxicity of organic solvents for fish trophic level. Toxicity was expressed as log(LC50), the concentration in mmol.L(-1) producing the 50% death of fish. The 141 chemically heterogeneous solvents of the dataset were described by physico-chemical descriptors and quantum theoretical parameters calculated via Density Functional Theory. The best subsets of solvent descriptors for LC50 prediction were chosen both through the Kubinyi function associated with Enhanced Replacement Method and a stepwise forward multiple linear regressions. The 4-parameters selected in the model were the octanol-water partition coefficient, LUMO energy, dielectric constant and surface tension. The predictive power and robustness of the QSAR developed were assessed by internal and external validations. Several techniques for training sets selection were evaluated: a random selection, a LC50-based selection, a balanced selection in terms of toxic and non-toxic solvents, a solvent profile-based selection with a space filling technique and a D-optimality onions-based selection. A comparison with fish LC50 predicted by ECOSAR model validated for neutral organics confirmed the interest of the QSAR developed for the prediction of organic solvent aquatic toxicity regardless of the mechanism of toxic action involved.

QSTR with extended topochemical atom (ETA) indices. 14. QSAR modeling of toxicity of aromatic aldehydes to Tetrahymena pyriformis

Aldehydes are a toxic class of chemicals causing severe health hazards. In this background, quantitative structure-toxicity relationship (QSTR) models have been developed in the present study using Extended Topochemical Atom (ETA) indices for a large group of 77 aromatic aldehydes for their acute toxicity against the protozoan ciliate Tetrahymena pyriformis. The ETA models have been compared with those developed using various non-ETA topological indices. Attempt was also made to include the n-octanol/water partition coefficient (logK(o/w)) as an additional descriptor considering the importance of hydrophobicity in toxicity prediction. Thirty different models were developed using different chemometric tools. All the models have been validated using internal validation and external validation techniques. The statistical quality of the ETA models was found to be comparable to that of the non-ETA models. The ETA models have shown the important effects of steric bulk, lipophilicity, presence of electronegative atom containing substituents and functionality of the aldehydic oxygen to the toxicity of the aldehydes. The best ETA model (without using logK(o/w)) shows encouraging statistical quality (Q(int)(2)=0.709,Q(ext)(2)=0.744). It is interesting to note that some of the topological models reported here are better in statistical quality than previously reported models using quantum chemical descriptors.