Comparison of In Silico Models for Prediction of Mutagenicity

Abstract

Using a dataset with more than 6000 compounds, the performance of eight quantitative structure activity relationships (QSAR) models was evaluated: ACD/Tox Suite, Absorption, Distribution, Metabolism, Elimination, and Toxicity of chemical substances (ADMET) predictor, Derek, Toxicity Estimation Software Tool (T.E.S.T.), TOxicity Prediction by Komputer Assisted Technology (TOPKAT), Toxtree, CEASAR, and SARpy (SAR in python). In general, the results showed a high level of performance. To have a realistic estimate of the predictive ability, the results for chemicals inside and outside the training set for each model were considered. The effect of applicability domain tools (when available) on the prediction accuracy was also evaluated. The predictive tools included QSAR models, knowledge-based systems, and a combination of both methods. Models based on statistical QSAR methods gave better results.

Using random forest and decision tree models for a new vehicle prediction approach in computational toxicology

“…Many useful computational models have been developed with varying success. Models ranging from studies on aquatic toxicity that utilise multiple linear regression and neural networks through to predictive hepatotoxicity (Cruz-Monteagudo et al 2008;Low et al 2011 and mutagenicity (Bakhtyari et al 2013;Xu et al 2012;Modi et al 2012) that use k-nearest neighbour, support vector machines and random forests are frequently reported in literature.…”

In Silico Prediction of Chemically Induced Mutagenicity: How to Use QSAR Models and Interpret Their Results

“…On the other hand, ADMET predictor (Absorption, Distribution, Metabolism, Elimination, and Toxicity of chemical substances), a commercial tool, includes ten different models for different strains of S. typhimurium with and without microsomal activation [ 29 ]. We notice that the performance of the "general" mutagenicity models was superior compared to the strain-specifi c models, when tested in a large set of compounds [ 20 ]. 3.…”

“…To compare the performance of three VEGA models, we applied them to the same evaluation set. This data set counts more than 6000 compounds evenly distributed between mutagens and nonmutagens and was used within the European LIFE project ANTARES for the evaluation of different QSAR models [ 20 ].…”

Molecular descriptors as proxies for the modeling of the materials and their environmental impact

“…Moreover the possibility to satisfactorily apply a model to a pharmaceutical of interest may depend on toxicological data availability for molecules chemically related to this entity. There are papers in which model performance is assessed for general chemicals [30], or for pharmaceuticals, with respect to carcinogenicity and mutagenicity [31]. In contrast, there is a paucity of models for developmental or reproductive toxicity [32].…”