The role of a molecular informatics platform to support next generation risk assessment

High Throughput Read-Across for Screening a Large Inventory of Related Structures by Balancing Artificial Intelligence/Machine Learning and Human Knowledge

“…We then further compared the molecular property profiles among the clusters. Within ChemTunes·ToxGPS, three types of molecular properties were employed in this study: (1) whole molecule, (2) shape, and (3) quantum mechanics-based energetic properties . Whole molecule properties are further divided into three groups to address bonding interactions, interfacial interactions and bioavailability, and molecular size.…”

“…For comparing the property profiles of structural clusters, we employed number of atoms, MW, rotational bonds, H-bond acceptors and donors, TPSA, XlogP, asphericity, Δ H f , HOMO, LUMO, and HOMO/LUMO. Plotting the variation of these patterns for each cluster resulted in distinct skyline profiles. , Skyline profiles are simple bar plots where each bar corresponds to a particular property standardized using the mean and standard deviation of that property calculated from a large data set of more than 50,000 representative structures from the ChemTunes database. These plots allow visual comparison of property profiles between different compounds.…”

“…Performing read-across quantitatively requires information about the studies with analogue data to rate these with appropriate study quality (SQ) values. SQ is one of the key parameters applied when determining the RR and analogue quality (AQ) . Several aspects of studies are evaluated, including objective study design-related parameters and more subjective reliability of results, which are part of the core information on the study inclusion criteria, namely, the COSMOS Minimum Inclusion (MINIS) grade developed in COSMOS project. , The critical studies used in the NOAEL data set had rigorously documented quality parameters and were assigned to five categories: High, Med-High, Med, Med-Low, and Low, which were numerically transformed to 1.0, 0.95, 0.90, 0.85, 0.75.…”

“…Hence, the algorithm to rate SQ systematically, developed in the COSMOS project, has been implemented in the ChemTunes NOAEL data set. For this reason, we simplified the “General Study Reliability Likelihood Based on MINIS Grade and Opinion Scores”, ,, reported previously so that the same process can be applied to NOAEL values needed in quantitative read-across (qRAX). Since PODs are determined from a critical toxicity study, in most cases, COSMOS MINIS grades for the study in ChemTunes can be used for the design and results parameters.…”

“…Also reported are the methods for quantification of the qualities of analogues and their study data, , which enables assessment of read-across reliabilities, outcomes (final resulting end point values, for example, an estimated point of departure (POD) for the target), and corresponding uncertainties. To support the needs, several public and commercial tools have become available in the past several years, including OECD QSAR Toolbox, VEGA HUB (including ToxREAD), ToxMatch, Generalized Read-Across GenRA , from the United States (US) Environmental Protection Agency (EPA) and ChemTunes·ToxGPS. , Several organizations use the latter platform as their internal system to support in silico safety/risk assessment, for example, the chemoinformatics group in the US Food and Drug Administration (FDA) Center for Food Safety and Applied Nutrition (CFSAN) and Cosmetics Europe. Feedback from these entities is further guiding the development of digitalized NGRA workflows, including approaches whereby analogue candidates are evaluated based on analogue quality (AQ) and study quality (SQ) metrics.…”