Abstract
SummaryComputational prediction of toxicity has reached new heights as a result of decades of growth in the magnitude and diversity of biological data. Public packages for statistics and machine learning make model creation faster. New theory in machine learning and cheminformatics enables integration of chemical structure, toxicogenomics, simulated and physical data in the prediction of chemical health hazards, and other toxicological information. Our earlier publications have characterized a toxicological dataset of unprecedented scale resulting from the European REACH legislation (Registration Evaluation Authorisation and Restriction of Chemicals). These publications dove into potential use cases for regulatory data and some models for exploiting this data. This article analyzes the options for the identification and categorization of chemicals, moves on to the derivation of descriptive features for chemicals, discusses different kinds of targets modeled in computational toxicology, and ends with a high-level perspective of the algorithms used to create computational toxicology models.
Highlights
The explosive growth of data from biology, physics and chemistry enables the creation and validation of powerful predictive models for toxicological endpoints
Since thiazolidenediones appear to be a relevant target of QSARs, we find the terms that are more prevalent for this group relative to all QSAR publications
In a study of the consistency of the Draize rabbit eye irritation test, we found a low rate of concordance between studies (Luechtefeld et al, 2016b)
Summary
The explosive growth of data from biology, physics and chemistry enables the creation and validation of powerful predictive models for toxicological endpoints. QSARs built on chemical similarity or “black box” machine learning methods such as neural networks do not provide a clean mechanistic interpretation In these latter models, statistical techniques are sometimes derived to help elucidate the modeled mechanisms (Matthews et al, 2009). To satisfy REACH and the parallel Classification, Labeling and Packaging (CLP) legislation, companies must generate United Nations Globally Harmonized System (GHS) classification and labeling for each chemical (Winder et al, 2005) These labels are alphanumeric identifiers (e.g., H317 = skin sensitization), which form ideal “well-defined” endpoints for QSARs. Depending on the chemical and its usage level, some substances can satisfy reporting requirements with QSARs and chemical similarity approaches (termed read-across) (Patlewicz et al, 2014; Ball et al, 2016). 7 http://www.oecd.org/env/ehs/risk-assessment/37849783.pdf (last accessed 11 Oct 2017)
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
