External validation of alert profilers for genotoxicity hazard within the OECD QSAR Toolbox
Genotoxic hazard identification is crucial for regulatory decision-making in many countries. Computational toxicity prediction serves as a useful first step in assessing pesticide impurities and metabolites. In addition to validated QSAR models, the OECD QSAR Toolbox offers mechanistic and endpoint-specific profiler information for chemical grouping, though their predictive performance varies by case. A study compared the genotoxicity-relevant profilers in the OECD QSAR Toolbox with experimental results from the CASE Ultra AMES mutagenicity and an in vivo MNT database to better understand their predictive accuracy and role in weight-of-evidence decisions for genotoxic endpoints.
‘Relevant’ versus ‘Similar’ Structural Analogs to Support Genotoxicity (Q)SAR Predictions
Structural analogs with experimental data are important for expert review of (Q)SAR model evaluations. Analogs help clarify uncertain (Q)SAR outcomes, making inconclusive or contradictory evaluations more definitive. Finding analogs is challenging when query compounds have multiple alerts of varied significance, a common scenario in statistical model evaluations. Conventional similarity metrics may yield structurally similar analogs but often lack relevance by failing to identify analogs with query compound’s alerts. Here we introduce a novel approach for finding analogs, employing a sophisticated fingerprinting technique that considers both specific alerts and general structural features.
Developing YosAI, an AI Genotoxicity Prediction System from Eisai
YosAI is an AI genotoxicity prediction system which integrates commercially available in silico genotoxicity prediction software with text mining technology based on Eisai’s ideas, structural feature data, and the knowledge and know-how of genotoxicity experts.
Evaluation of QSAR models for predicting mutagenicity: outcome of the Second Ames/QSAR international challenge project
To promote superior model development, the Division of Genetics and Mutagenesis, National Institute of Health Sciences, Japan (DGM/NIHS), conducted the Second Ames/QSAR International Challenge Project.
Augmenting Expert Knowledge-Based Toxicity Alerts by Statistically Mined Molecular Fragments
In this work, we present a method to build hybrid QSAR models by combining expert knowledge-based alerts and statistically mined molecular fragments.
Reason Vectors: Abstract Representation of Chemistry–Biology Interaction Outcomes, for Reasoning and Prediction
We developed a method for learning higher-level abstract representations of the effects of the interactions between molecular features and biology.
Descriptor Free QSAR Modeling Using Deep Learning with Long Short-Term Memory Neural Networks
In this study, we explored the prospects of building good quality interpretable QSARs for big and diverse datasets, without using any pre-calculated descriptors.
Effectiveness of CASE Ultra Expert System in Evaluating Adverse Effects of Drugs
The purpose of this pilot study is to test the QSAR expert system CASE Ultra for adverse effect prediction of drugs.