You are here:
Evaluation of the Chemotype-Enrichment Workflow: A Tool for Independent Evaluation of Biological Activity Thresholds and a Comparison with QSAR Methods
Citation:
Lougee, R., A. Richard, Chris Grulke, AND A. Williams. Evaluation of the Chemotype-Enrichment Workflow: A Tool for Independent Evaluation of Biological Activity Thresholds and a Comparison with QSAR Methods. Presented at American Chemical Society Spring Meeting, Orlando, FL, March 31 - April 04, 2019. https://doi.org/10.23645/epacomptox.7973348
Impact/Purpose:
In our first case study, we employed the CTEW to examine the overall impact on CTEW enrichment results when analyzing ToxCast assay results with and without application of a biological cytotoxicity filter. Cross validation is used to evaluate the cytoxicity filter and non-filtered results across the ToxCast Assays, and examine their activity against a random sample. Secondly, to bridge our understanding of the benefits and limitations of the CTEW approach in relation to global QSAR modeling, we compared balanced accuracy statistics of the full set of enriched CTs for each ToxCast assay dataset with the results from gradient-boosted random forest QSAR models applied to the same ToxCast assay datasets.
Description:
The Chemotype-Enrichment Workflow (CTEW), an automated univariate analysis tool, was developed to explore activity enrichments within local chemistry domains associated with EPA’s ToxCast bioassay results. The current implementation of the CTEW employs the publicly available ToxPrint chemotype (CT) set to define a fixed chemical abstraction layer for use in exploring individual assay datasets, as well as for comparing enrichment results globally, across assay endpoints and datasets. For each assay, the CTEW processes an input file consisting of a list of structures and binary hit-calls [1,0], generates CT fingerprints, and applies a set of defined statistical thresholds to identify CT enrichments in positive [1] and negative [0] activity space. ToxPrints offer a set of visualizable chemical features that can facilitate interpretation of results and support hypothesis generation. CTEW enrichment results can be considered an “attribute” of each assay dataset(s) and, as such, can be used to probe biological activity threshold assumptions (usually expert-based) that define the associated hit-calls. In addition, CTEW results can identify local activity enrichments in small, noisy, or imbalanced datasets that are less amenable to global Quantitative Structure-Activity Relationship (QSAR) analysis. In our first case study, we employed the CTEW to examine the overall impact on CTEW enrichment results when analyzing ToxCast assay results with and without application of a biological cytotoxicity filter. Cross validation is used to evaluate the cytoxicity filter and non-filtered results across the ToxCast Assays, and examine their activity against a random sample. Secondly, to bridge our understanding of the benefits and limitations of the CTEW approach in relation to global QSAR modeling, we compared balanced accuracy statistics of the full set of enriched CTs for each ToxCast assay dataset with the results from gradient-boosted random forest QSAR models applied to the same ToxCast assay datasets. This abstract does not reflect U.S. EPA policy.
URLs/Downloads:
DOI: Evaluation of the Chemotype-Enrichment Workflow: A Tool for Independent Evaluation of Biological Activity Thresholds and a Comparison with QSAR Methods
LOUGEE_ACS_ORLANDO_PRES_2019 AR2.PDF (PDF, NA pp, 13805.422 KB, about PDF)