You are here:
An Evaluation of Performance and Coverage of Selected in Silico Metabolism Tools
Citation:
Boyce, M., B. Meyer, Chris Grulke, AND G. Patlewicz. An Evaluation of Performance and Coverage of Selected in Silico Metabolism Tools. ASCCT, RTP, NC, October 20 - 23, 2020. https://doi.org/10.23645/epacomptox.24290584
Impact/Purpose:
Societal demands for safer and sustainable chemical products are stimulating changes in toxicity testing and assessment frameworks. Chemical safety assessments are expected to be conducted faster and with fewer animals, yet the number of chemicals that require assessment is also rising with the number of different regulatory programmes worldwide. Read-across is a technique for identifying similar chemicals (or analogues) in an attempt to evaluate whether the toxicities and concerns associated with those analogues should be considered a potential liability for the chemical of interest. This research area will further develop quantitative, objective, flexible and transparent, tools and approaches to assess chemical and biological similarity, and to quantify uncertainties in predictions. The results of these efforts will be of direct benefit to program and regional offices as well as the greater scientific community.
Description:
Regulatory changes in chemical safety assessment have sought to supplant animal testing with high-throughput screening and in silico modeling. In conjunction with these regulatory changes, metabolism prediction software have seen significant growth in recent years and can support risk assessment by identifying potentially toxic metabolites. As the number of commercially and freely available metabolism prediction tools continue to grow, performance comparisons are needed to provide context for an informed selection. Here we compiled a reference set of 37 structurally diverse chemicals from the EPA ExpoCast inventory to evaluate the performance and coverage of a selection of in silico models: Systematic Generation of Metabolites (SyGMa), Meteor Nexus, BioTransformer, Tissue Metabolism Simulator (TIMES), and OECD Toolbox. Structural information for the reference chemicals were retrieved from the CompTox Chemicals Dashboard as SMILES. Predictions were run as batches using default settings provided by software (Meteor, TIMES, Toolbox). In cases where users must define metabolism schemes (e.g., SyGMa and BioTransfomer), two passes of phase I and one pass of phase II were used. Inter-model comparisons found TIMES and OECD Toolbox to be the most similar. SyGMa had the greatest coverage, matching an average of 41.2% of predictions generated by the other models. However, SyGMa was also prone to significant overpredicting, generating a total of 5,125 predictions or 67% of total predictions. Current efforts are focused on evaluating the precision and sensitivity of the models by comparing their predictions against metabolites reported in literature and empirical measurements. This abstract does not reflect EPA policy.
URLs/Downloads:
DOI: An Evaluation of Performance and Coverage of Selected in Silico Metabolism Tools
POSTER.PDF (PDF, NA pp, 598.183 KB, about PDF)