Making Negative Predictions for Mutagenicity.pdf
Derek Nexus contains new expert-derived functionality to provide negative predictions for bacterial in vitro mutagenicity. Nonalerting compounds are evaluated to identify unclassified and misclassified features (from a data set of >104 compounds).
Misclassified features in the molecule are derived from nonalerting mutagens in the Lhasa reference set.
Features in the molecule that are not found in the Lhasa reference set are considered unclassified.
In compounds where all features in the molecule are found in accurately classified compounds from the reference set, a negative prediction is displayed. Predictions for compounds with misclassified or unclassified features remain negative, and these features are highlighted to the user to enable expert assessment of the prediction.
Performance against three external, proprietary data sets highlights that negative predictivity for all outcomes is good (generally > 80%) with the presence of unclassified or misclassified features slightly reducing accuracy.
Presented by Richard Williams at GTA meeting, Delaware, USA; 7th May 2014.