Derek Nexus – Achieving high accuracy with high coverage [an infographic]
11 August 2021
Read time: less than 5 minutes
In this blog piece, we explore how Derek Nexus, our expert, knowledge-based toxicology software, performs against 7 alternative in silico skin sensitisation models.
Providing qualitative predictions for many toxicity endpoints, including assessment of skin sensitisation, Derek can predict quantitative EC3 predictions for skin sensitisation as well as negative predictions for those query compounds which do not fire a skin sensitisation alert.
When evaluated against 7 different in silico skin sensitisation models, Derek achieves high accuracy with high coverage. A recent publication, 'Evaluation of the Global Performance of Eight in silico Skin Sensitization Models Using Human Data', analyses the accuracy of eight in silico skin sensitisation models against two human data sets: one highly curated (Basketter et al., 2014) and one screening level (Hazardous Substances Data Bank).
The binary skin sensitisation status of each chemical in both data sets was compared to the prediction from eight in silico skin sensitisation tools (Toxtree, PredSkin, OECD’s QSAR Toolbox, UL’s REACHAcross™, Danish QSAR Database, TIMES-SS and Derek Nexus). Models were assessed for coverage, accuracy, sensitivity and specificity, as well as optimisation features (e.g. probability of accuracy, applicability domain, etc.), if available.
Derek performs well for both Basketter et al. 2014 and HSDB data sets, combining high coverage (98% and 90% respectively) with high accuracy (86% and 70% respectively) when using base settings.
See the infographic below, to discover how Derek measured up against the other in silico tools.
About this article
Authorby Emma Scrafton
less than 5 minutes