Using probabilistic neural networks to model the toxicity of chemicals to the fathead minnow (Pimephales promelas): A study based on 865 compounds

doi:10.1016/S0045-6535(99)00553-6

Chemosphere

Volume 38, Issue 14, June 1999, Pages 3237-3245

https://doi.org/10.1016/S0045-6535(99)00553-6 Get rights and content

Abstract

We investigate the use of probabilistic neural networks (PNN) to model the acute toxicity (96-hr LC50) to the fathead minnow (Pimephales promelas) based on a 865 chemicals data set. In contrast to most other toxicological models, the octanol/water partition coefficient is not used as input parameter. The information fed into the neural network is solely based on simple molecular descriptors as can be derived from the chemicals' structures and indicates the potential of this approach as general methodology for the estimation of toxicological effects of chemicals.

References (13)

R.D. Cramer et al.
Crossvalidation, bootstrapping, and partial least squares compared with multiple regression in conventional QSAR studies
Quantitative Structure-Activity Relationships
(1988)
K.L.E. Kaiser et al.
Neural network modeling of Vibrio fischeri toxicity data with structural physico-chemical parameters and molecular indicator variables. Poster
K.L.E. Kaiser et al.
Neural network modeling of Vibrio fischeri and fathead minnow acute toxicity data with molecular indicator variables and physico-chemical bulk parameters. Poster

There are more references available in the full text version of this article.

Cited by (41)

Combination of high resolution mass spectrometry and a halogen extraction code to identify chlorinated disinfection byproducts formed from aromatic amino acids
2021, Water Research
Citation Excerpt :
The prediction model, acute toxicity (96-h half-maximal lethal concentration [LC50]) to the fathead minnow, was used to assess the comparative toxicity of Cl-DBPs. Detailed information regarding toxicity predictions has also been reported previously (Eldred et al., 1999; Kaiser and Niculescu, 1999; Karabunarliev et al., 1996; Konemann, 1981; Martin and Young, 2001; Nendza and Russom, 1991; Newsome et al., 1991; Russom et al., 1997; Zhu et al., 2009). The dried organic byproducts obtained by solid-phase extraction were dissolved in methanol/ultrapure water (1 mL, 1:1 [v/v]).
Chlorination can lead to the formation of hazardous chlorinated disinfection byproducts (Cl-DBPs). We identified tyrosine (Tyr) and tryptophan (Trp) as precursors of toxic Cl-DBPs and developed a halogen extraction code to complement ultra performance liquid chromatography in tandem with high resolution mass spectrometry (UPLC-HRMS) in detecting and identifying Cl-DBPs. We detected 20 and 11 Cl-DBPs formed from chlorination of Tyr and Trp, respectively, and identified the structures of 15 Cl-DBPs. Fourteen structures were previously unreported. We also proposed the tentative formation pathways of these newly identified Cl-DBPs. Their incidence in real water sources demonstrated that these Cl-DBPs are likely to form during chlorination of reclaimed water. We computationally predicted the toxicity of these Cl-DBPs, which was relatively high, indicating that these Cl-DBPs could be hazardous and were of valid concern. Combining analytical data with the halogen extraction code can identify Cl-DBPs accurately from complex compounds. This analytical method can be used to identify Cl-DBPs of water treatment procedures in further studies.
In silico prediction of toxicity of phenols to Tetrahymena pyriformis by using genetic algorithm and decision tree-based modeling approach
2017, Chemosphere
Citation Excerpt :
Possibility to overfit a dataset with too many nonbiologically relevant chemical descriptors is the disadvantage of global methods. A number of methods such as multiple linear regression (MLR) (Su et al., 2012; Tugcu et al., 2012), group contribution methods (Martin and Young, 2001), partial least squares (Castillo-Garit et al., 2008), artificial neural networks (Eldred et al., 1999; Kaiser and Niculescu, 1999), k nearest neighbors (Cassotti et al., 2014), and ensemble learning approach (Singh and Gupta, 2014) were used for the QSAR modeling. These models have some drawbacks that can limit their application for regulatory purposes.
Risk assessment of chemicals is an important issue in environmental protection; however, there is a huge lack of experimental data for a large number of end-points. The experimental determination of toxicity of chemicals involves high costs and time-consuming process. In silico tools such as quantitative structure–toxicity relationship (QSTR) models, which are constructed on the basis of computational molecular descriptors, can predict missing data for toxic end-points for existing or even not yet synthesized chemicals. Phenol derivatives are known to be aquatic pollutants. With this background, we aimed to develop an accurate and reliable QSTR model for the prediction of toxicity of 206 phenols to Tetrahymena pyriformis. A multiple linear regression (MLR)-based QSTR was obtained using a powerful descriptor selection tool named Memorized_ACO algorithm. Statistical parameters of the model were 0.72 and 0.68 for $R_{t r a i n i n g}^{2}$ and $R_{t e s t}^{2}$ , respectively. To develop a high-quality QSTR model, classification and regression tree (CART) was employed. Two approaches were considered: (1) phenols were classified into different modes of action using CART and (2) the phenols in the training set were partitioned to several subsets by a tree in such a manner that in each subset, a high-quality MLR could be developed. For the first approach, the statistical parameters of the resultant QSTR model were improved to 0.83 and 0.75 for $R_{t r a i n i n g}^{2}$ and $R_{t e s t}^{2}$ , respectively. Genetic algorithm was employed in the second approach to obtain an optimal tree, and it was shown that the final QSTR model provided excellent prediction accuracy for the training and test sets ( $R_{t r a i n i n g}^{2}$ and $R_{t e s t}^{2}$ were 0.91 and 0.93, respectively). The mean absolute error for the test set was computed as 0.1615.
The proposal of architecture for chemical splitting to optimize QSAR models for aquatic toxicity
2008, Chemosphere
One of the challenges in the field of quantitative structure-activity relationship (QSAR) analysis is the correct classification of a chemical compound to an appropriate model for the prediction of activity. Thus, in previous studies, compounds have been divided into distinct groups according to their mode of action or chemical class. In the current study, theoretical molecular descriptors were used to divide 568 organic substances into subsets with toxicity measured for the 96-h lethal median concentration for the Fathead minnow (Pimephales promelas). Simple constitutional descriptors such as the number of aliphatic and aromatic rings and a quantum chemical descriptor, maximum bond order of a carbon atom divide compounds into nine subsets. For each subset of compounds the automatic forward selection of descriptors was applied to construct QSAR models. Significant correlations were achieved for each subset of chemicals and all models were validated with the leave-one-out internal validation procedure $(R_{cv}^{2} \approx 0.80)$ . The results encourage to consider this alternative way for the prediction of toxicity using QSAR subset models without direct reference to the mechanism of toxic action or the traditional chemical classification.
Algorithms for (Q)SAR model building
2007, Quantitative Structure-Activity Relationships (QSAR) for Pesticide Regulatory Purposes
Assessing the reliability of a QSAR model's predictions
2005, Journal of Molecular Graphics and Modelling
Quantitative structure activity relationships (QSAR) are one of the well-developed areas in computational chemistry. In this field, many successful predictive models have been developed for various property, activity or toxicity predictions. However, the predictive power of models for new query compounds is often not well characterized. The breadth of applicability of models is often not characterized. In other words, with a given QSAR model and a specific query compound to be predicted, can the model be used reliably for the desired prediction? In this study, we assessed the reliability of QSAR models’ prediction on query compounds. Our approach, employing hierarchical clustering, was developed and tested using a test dataset containing 322 organic compounds with fathead minnow acute aquatic toxicity as the activity of interest. The hypothesis of the approach was that if a query compound is more similar to the compounds used to generate the QSAR model, it should be predicted more accurately. Thus, the core of the approach is to determine the relationship between the similarity of query compounds to the training set compounds of the QSAR model and the prediction accuracy given by that model. This relationship determination was achieved by comparing the results given by the two major components of the approach: objects clustering and activity prediction. With the resultant information from the two steps, a direct relationship was shown.
A novel approach to predict a toxicological property of aromatic compounds in the Tetrahymena pyriformis
2004, Bioorganic and Medicinal Chemistry
The TOPological Substructural MOlecular DEsign (TOPS-MODE) has been successfully used in order to explain the toxicity in the Tetrahymena pyriformis on a large data set. The obtained models for the training set had good statistical parameters (R²=0.72–0.81, p<0.05) an also the prediction power of the models found was adequate (Q²=0.70–0.80). A detailed study of the influence of variable numbers in the equation and the statistical outliers was carried out; leading to a good final model with a better physicochemical interpretation than the rest of the published models. Only two molecular descriptors codifying dipolar and hydrophobic features were introduced. Finally, the fragment contributions to the toxicity prediction evidenced the powerful of this topological approach.

View all citing articles on Scopus

View full text

Article preview

Chemosphere

Abstract

References (13)

Crossvalidation, bootstrapping, and partial least squares compared with multiple regression in conventional QSAR studies

Quantitative Structure-Activity Relationships

Neural network modeling of Vibrio fischeri toxicity data with structural physico-chemical parameters and molecular indicator variables. Poster

Neural network modeling of Vibrio fischeri and fathead minnow acute toxicity data with molecular indicator variables and physico-chemical bulk parameters. Poster

Cited by (41)

Combination of high resolution mass spectrometry and a halogen extraction code to identify chlorinated disinfection byproducts formed from aromatic amino acids

In silico prediction of toxicity of phenols to Tetrahymena pyriformis by using genetic algorithm and decision tree-based modeling approach

The proposal of architecture for chemical splitting to optimize QSAR models for aquatic toxicity

Algorithms for (Q)SAR model building

Assessing the reliability of a QSAR model's predictions

A novel approach to predict a toxicological property of aromatic compounds in the Tetrahymena pyriformis