Alert performance: A new functionality in the OECD QSAR Toolbox

Yordanova, Darina; Schultz, T.W.; Kuseva, Chanita; Ivanova, H.; Pavlov, Todor; Chankov, Georgi; Karakolev, Yordan H.; Gissi, Andrea; Sobański, Tomasz; Mekenyan, Ovanes G.

doi:10.1016/j.comtox.2018.12.003

Cited by 14 publications

(6 citation statements)

References 7 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This end point is interesting for benchmarking, as the model is required to learn a wide range of different chemical features. Although the data set used is proprietary, publicly available toxicity alerts (e.g., ToxAlerts, OECD QSAR Toolbox) , may be leveraged as useful public benchmark data sets.…”

Section: Discussionmentioning

confidence: 99%

Interpreting Neural Network Models for Toxicity Prediction by Extracting Learned Chemical Features

Walter,

Webb,

Gillet

2024

J. Chem. Inf. Model.

View full text Add to dashboard Cite

Neural network models have become a popular machine-learning technique for the toxicity prediction of chemicals. However, due to their complex structure, it is difficult to understand predictions made by these models which limits confidence. Current techniques to tackle this problem such as SHAP or integrated gradients provide insights by attributing importance to the input features of individual compounds. While these methods have produced promising results in some cases, they do not shed light on how representations of compounds are transformed in hidden layers, which constitute how neural networks learn. We present a novel technique to interpret neural networks which identifies chemical substructures in training data found to be responsible for the activation of hidden neurons. For individual test compounds, the importance of hidden neurons is determined, and the associated substructures are leveraged to explain the model prediction. Using structural alerts for mutagenicity from the Derek Nexus expert system as ground truth, we demonstrate the validity of the approach and show that model explanations are competitive with and complementary to explanations obtained from an established feature attribution method.

show abstract

Section: Discussionmentioning

confidence: 99%

Interpreting Neural Network Models for Toxicity Prediction by Extracting Learned Chemical Features

Walter,

Webb,

Gillet

2024

J. Chem. Inf. Model.

View full text Add to dashboard Cite

show abstract

“…By applying the profilers (profiling) to the chemicals of interest, Toolbox is able to determine their structural and mechanistic characteristics based on the available information in the used profilers. Thus, the profiling information for a chemical could be used to search for analogues, to remove dissimilar chemicals and, accordingly, to reduce the uncertainty of the category‐based predictions (Yordanova, Kusev, et al, 2019; Yordanova, Schultz, et al, 2019). In some of the profilers containing endpoint‐specific knowledge, this knowledge is supported by relevant training set chemicals.…”

Section: Methodsmentioning

confidence: 99%

Selection of Representative Constituents for Unknown, Variable, Complex, or Biological Origin Substance Assessment Based on Hierarchical Clustering

Yordanova

Patterson

North

et al. 2021

Enviro Toxic and Chemistry

Self Cite

View full text Add to dashboard Cite

Many of the newly produced and registered substances are complex mixtures or substances of unknown or variable composition, complex reaction products, and biological materials (UVCBs). The latter often consist of a large number of constituents, some of them difficult‐to‐identify constituents, which complicates their (eco)toxicological assessment. In the present study, through a series of examples, different scenarios for selection of representatives via hierarchical clustering of UVCB constituents are exemplified. Hierarchical clustering allows grouping of the individual chemicals into small sets, where the constituents are similar to each other with respect to more than one criterion. To this end, various similarity criteria and approaches for selection of representatives are developed and analyzed. Two types of selection are addressed: (1) selection of the most “conservative” constituents, which could be also used to support prioritization of UVCBs for evaluation, and (2) obtaining of a small set of chemical representatives that covers the structural and metabolic diversity of the whole target UVCBs or a mixture that can then be evaluated for their environmental and (eco)toxicological properties. The first step is to generate all plausible UVCB or mixture constituents. It was found that the appropriate approach for selecting representative constituents depends on the target endpoint and physicochemical parameters affecting the endpoint of interest. Environ Toxicol Chem 2021;40:3205–3218. © 2021 SETAC

show abstract

“…Even then, the best diagnostic tools will have little value unless they are used by operators who are familiar with the RAx technique as well as the underlying toxicology and chemistry of the specific case and the context-specific criteria for scientific validity. For example, the most recent version of the OECD QSAR Toolbox 16 supports the user by providing information on alert performance based on the category candidates that are selected ( Yordanova et al, 2019 ). Recently, Luechtefeld and co-workers (2018a) attempted this task using an automated RAx-based QSAR approach on a huge set of acute and topical toxicity as well as ecotoxicological endpoints.…”

Section: Regulatory Use Of Raxmentioning

confidence: 99%

Internationalization of read-across as a validated new approach method (NAM) for regulatory toxicology

Rovida

Barton-Maclaren

Benfenati

et al. 2020

ALTEX

View full text Add to dashboard Cite

dossier to the European Chemicals Agency (ECHA) containing an extensive list of data on the intrinsic properties of a substance, ranging from chemical characterization and physicochemical properties to toxicological and ecotoxicological data, with increasing demands depending on the tonnage band of the quantity of the substance placed on the market. In addition, registrants should collect information on use and exposure to perform a chemical safety assessment (CSA) based on the toxicity profile of the substance. The minimum data requirements are described in Annexes VI-X of the regulation. For toxicological testing, the standard require-

show abstract

Alert performance: A new functionality in the OECD QSAR Toolbox

Cited by 14 publications

References 7 publications

Interpreting Neural Network Models for Toxicity Prediction by Extracting Learned Chemical Features

Interpreting Neural Network Models for Toxicity Prediction by Extracting Learned Chemical Features

Selection of Representative Constituents for Unknown, Variable, Complex, or Biological Origin Substance Assessment Based on Hierarchical Clustering

Internationalization of read-across as a validated new approach method (NAM) for regulatory toxicology

Contact Info

Product

Resources

About