Andrey A. Toropov

Using nano-QSAR to predict the cytotoxicity of metal oxide nanoparticles

It is expected that the number and variety of engineered nanoparticles will increase rapidly over the next few years, and there is a need for new methods to quickly test the potential toxicity of these materials. Because experimental evaluation of the safety of chemicals is expensive and time-consuming, computational methods have been found to be efficient alternatives for predicting the potential toxicity and environmental impact of new nanomaterials before mass production. Here, we show that the quantitative structure-activity relationship (QSAR) method commonly used to predict the physicochemical properties of chemical compounds can be applied to predict the toxicity of various metal oxides. Based on experimental testing, we have developed a model to describe the cytotoxicity of 17 different types of metal oxide nanoparticles to bacteria Escherichia coli. The model reliably predicts the toxicity of all considered compounds, and the methodology is expected to provide guidance for the future design of safe nanomaterials.

CORAL: Quantitative structure–activity relationship models for estimating toxicity of organic compounds in rats

For six random splits, one‐variable models of rat toxicity (minus decimal logarithm of the 50% lethal dose [pLD50], oral exposure) have been calculated with CORAL software (http://www.insilico.eu/coral/). The total number of considered compounds is 689. New additional global attributes of the simplified molecular input line entry system (SMILES) have been examined for improvement of the optimal SMILES‐based descriptors. These global SMILES attributes are representing the presence of some chemical elements and different kinds of chemical bonds (double, triple, and stereochemical). The “classic” scheme of building up quantitative structure–property/activity relationships and the balance of correlations (BC) with the ideal slopes were compared. For all six random splits, best prediction takes place if the aforementioned BC along with the global SMILES attributes are included in the modeling process. The average statistical characteristics for the external test set are the following: n = 119 ± 6.4, R2 = 0.7371 ± 0.013, and root mean square error = 0.360 ± 0.037.

Novel application of the CORAL software to model cytotoxicity of metal oxide nanoparticles to bacteria Escherichia coli.

Convenient to apply and available on the Internet software CORAL (http://www.insilico.eu/CORAL) has been used to build up quantitative structure-activity relationships (QSAR) for prediction of cytotoxicity of metal oxide nanoparticles to bacteria Escherichia coli (minus logarithm of concentration for 50% effect pEC50). In this study six random splits of the data into the training and test set were examined. It has been shown that the CORAL provides a reliable tool that could be used to build up a QSAR of the pEC50.

Prediction of heteroaromatic amine mutagenicity by means of correlation weighting of atomic orbital graphs of local invariants

Abstract Graphs of atomic orbitals (GAOs) have been used to represent molecular structures. Rules by which the labeled hydrogen-filled graphs (LHFGs) were converted into the GAOs are described. The GAO is an attempt at taking into account the structures of atoms (i.e. atomic orbitals, such as 1s 1 , 2p 2 , 3d 10 ) for QSPR/QSAR analyses. As a method of mutagenicity modeling, optimization of correlation weights of local invariants (OCWLI) of the LHFGs and the GAOs has been used. Statistical characteristics of such models based on the OCWLI of GAO are better than those based on the OCWLI of the LHFGs.

QSAR modeling of acute toxicity by balance of correlations

Optimal descriptors based on the simplified molecular input line entry system (SMILES) have been utilized in modeling of acute toxicity towards rats. Toxicity of 61 benzene derivatives has been modeled by means of balance of correlations for sets of the training (n=27) and calibration (n=24). The obtained models were evaluated with the external test set (n=10). Comparison of models based on the balance of correlations and models which were obtained on base of the total training (i.e., in case of utilization both training and calibration sets as the united training set) has shown that the balance of correlations gives improvement of statistical quality for the external test set. Predictions based on the one-variable model (based on the correlation balance) are better that the results obtained by the multiple linear regression analysis based on topological and quantum chemical descriptors. A QSAR analysis showed that the electronegativity of the molecule plays an important role in acute toxicity of benzene derivatives studied; presence of electronegative groups increasing toxicity. The presence of nitrogen-containing groups (mostly NH groups) increasing the toxicity that confirmed by both approaches.

QSAR modeling of toxicity on optimization of correlation weights of Morgan extended connectivity

Abstract Optimal descriptors calculated in the presence of correlation weights in molecular graph of different kinds of vertices (i.e. chemical elements, e.g. H, C, O, etc.) and different Morgan extended connectivity of third-order values are reasonably better models of the acute aquatic toxicity (−log[LC 50 ]) of benzene derivatives. Statistical characteristics of the model are the following: on training set n =44, r 2 =0.8982, s =0.251, F =371; on test set n =25, r 2 =0.9181, s =0.234, F =258.

Quasi-QSAR for mutagenic potential of multi-walled carbon-nanotubes

Available on the Internet, the CORAL software (http://www.insilico.eu/coral) has been used to build up quasi-quantitative structure-activity relationships (quasi-QSAR) for prediction of mutagenic potential of multi-walled carbon-nanotubes (MWCNTs). In contrast with the previous models built up by CORAL which were based on representation of the molecular structure by simplified molecular input-line entry system (SMILES) the quasi-QSARs based on the representation of conditions (not on the molecular structure) such as concentration, presence (absence) S9 mix, the using (or without the using) of preincubation were encoded by so-called quasi-SMILES. The statistical characteristics of these models (quasi-QSARs) for three random splits into the visible training set and test set and invisible validation set are the following: (i) split 1: n=13, r(2)=0.8037, q(2)=0.7260, s=0.033, F=45 (training set); n=5, r(2)=0.9102, s=0.071 (test set); n=6, r(2)=0.7627, s=0.044 (validation set); (ii) split 2: n=13, r(2)=0.6446, q(2)=0.4733, s=0.045, F=20 (training set); n=5, r(2)=0.6785, s=0.054 (test set); n=6, r(2)=0.9593, s=0.032 (validation set); and (iii) n=14, r(2)=0.8087, q(2)=0.6975, s=0.026, F=51 (training set); n=5, r(2)=0.9453, s=0.074 (test set); n=5, r(2)=0.8951, s=0.052 (validation set).

Additive SMILES-based optimal descriptors in QSAR modelling bee toxicity: Using rare SMILES attributes to define the applicability domain.

The additive SMILES-based optimal descriptors have been used for modelling the bee toxicity. The influence of relative prevalence of the SMILES attributes in a training and test sets to the models for bee toxicity has been analysed. Avoiding the use of rare attributes improves statistical characteristics of the model on the external test set. The possibility of using the probability of the presence of SMILES attributes in training and test sets for rational definition of the applicability domain is discussed.

SMILES in QSPR/QSAR Modeling: Results and Perspectives

The technique of constructing optimal descriptors calculated with the Simplified molecular input line entry system (SMILES) is described. SMILES based optimal descriptors and descriptors calculated with molecular graphs (hydrogen filled graphs and graph of atomic orbitals) are compared in modeling done by means of quantitative structure - property/activity relationships (QSPR/QSAR). QSPR/QSAR models for normal boiling points of organic compounds, mutagenicity of heteroaromatic amines, toxicity, and anti-HIV-1 potentials of TIBO and HEPT derivatives are described in details. Possible ways to improve the SMILES based concept of QSPR/QSAR analyses are discussed.

Prediction of Aquatic Toxicity: Use of Optimization of Correlation Weights of Local Graph Invariants

Quantitative structure-activity relationships (QSARs) were developed for three sets of toxicity data. Chemicals in each set represented a number of narcoses and electrophilic mechanisms of toxic action. A series of quantitative structure-toxicity models correlating toxic potency with a number of optimization of correlation weights of local graph invariants were developed. In the case of the toxicity of a heterogeneous set of benzene derivatives to Tetrahymena pyriformis, the QSARs were based on the Descriptor of Correlation Weights (DCW) using atoms and extended connectivity (EC) graph invariants. The model [log (IGC(50)(-1)) = 0.0813 DCW(a(k),(3)EC(k)) + 2.636; n = 157, r(2) = 0.883, s = 0.27, F = 1170, Pr > F = 0.0001] based on third-order EC of 89 descriptors was observed to be best for the benzene data. However, fits for these data of > 0.800 were achieved ECs with as few as 23 variables. The relationship between the toxicity predicted by this model and experimental toxicity values for the test set [obs. log(IGC(50)(-1))) = 0.991 (pred. (log(IGC(50)(-1))) - 0.012; n = 60, r(2) = 0.863, s = 0.28, F = 372, Pr > F = 0.0001] is excellent. The utility of the approach was demonstrated by the model [log (IGC(50)(-1)) = 0.1744(DCW (a(k), (2)EC) - 3.505; n = 39, r(2) = 0.900, s = 0.35, F = 333, Pr > F = 0.0001] for the toxicity data for T. pyriformis exposed to halo-substituted aliphatic compounds and the model [log (IC(50)(-1)) = 0.1699(DCW (a(k), (2)EC)) - 2.610; n = 66, r(2) = 0.901, s = 0.31, F = 583, Pr > F = 0.0001] for the Vibrio fischeri toxicity data.