Automated and standardized workflows in the OECD QSAR Toolbox
Computational Toxicology Volume 10 (2019) 89-104
Toxicological understanding and computational possibilities progress every year. The OECD QSAR Toolbox, an in silico system for predicting toxicity of chemicals, is regularly updated with up to data knowledge and functionalities. This expansion has made the Toolbox increasingly complex and difficult to use for non-experts. Recent activities addressed this issue and enhanced the user-friendliness of the system. In a streamlining exercise, we developed automated and standardized workflows for acute aquatic toxicity and skin sensitization. These endpoints have been chosen based on understanding of the mechanisms of action, availability of experimental data and regulatory relevance. The workflows are designed to produce rational and reliable predictions, or no prediction if acceptance criteria are not met. The automated workflow is a predefined algorithm that generates a prediction without further interaction with the user after activation. The standardised workflow uses a similar algorithm but leaves to the user the key choices among a selection of suitable options. In the present work, we described in detail the algorithms for both the automated and standardized workflows. Using examples, the rationale behind each step of the two processes is explained.