Science Inventory

A Benchmark Concentration Analysis Method for Zebrafish Larval Locomotor Response Data Using ToxCast Pipeline Software.

Citation:

Rowson, Z., Woodrow Setzer, K. Paul-Friedman, R. Judson, AND S. Padilla. A Benchmark Concentration Analysis Method for Zebrafish Larval Locomotor Response Data Using ToxCast Pipeline Software. Society of Toxicology (FutureTox V), Chapel Hill, NC, May 10 - 11, 2022. https://doi.org/10.23645/epacomptox.19692406

Impact/Purpose:

Poster presented to the FutureTox V organ-specific toxicology Meeting May 2022. This work applies ToxCast's tcplfit2 curve-fitting software to zebrafish locomotor response data while simultaneously reducing the information lost from dimension reduciton of the data in to endpoints. This work prepares data for integration into the public data resource, ToxCast, supporting the integration of New Approach Methodologies for developmental neurotoxicity screening.

Description:

Larval zebrafish (Danio rerio) locomotor response (LMR) behavior may inform developmental neurotoxicity hazard, but there is no widely accepted analysis method for the high-throughput zebrafish LMR assay. This work proposes a method for evaluating chemical activity in the LMR assay that produces probabilistic representations of the likelihood of chemical activity and estimates point of departure (POD) using Benchmark Concentration (BMC) analysis. These metrics can be used to group chemicals by their activity and potency to elucidate common behavioral phenotypes. To allow for BMC analysis, complex time-series data derived from zebrafish larvae exposed during development to 61 chemicals were summarized into 16 endpoints. These endpoints characterized larval activity during the LMR: average speed, acceleration, and jerk; area under the curve; and habituation in Light or Dark, as well as acceleration during Light/Dark transition startles. After Box-Cox power transformation, curves were fit using the new ToxCast Pipeline curve-fitting and hit-calling utility, tcplfit2, a publicly available R package. Chemicals were grouped by their activity in each endpoint and clustered by their activity across all 16 endpoints. The analysis approach was compiled into a new R package, gabi, which summarized and transformed longitudinal data prior to application of tcplfit2. Of the 61 assayed chemicals, 22 were found to be active in at least one endpoint.  No more than 7 chemicals were active in any given endpoint and all endpoints other than those describing jerk in Light or Dark found at least two chemicals to be active. The sets of active chemicals associated with these endpoints were often unique, implying that endpoints detected different kinds of behavioral change. Characterization of chemical activity in multiple endpoints better captured changes in the LMR that were implied by plots of the longitudinal data.  This analysis method quantifies the significance of behavioral changes and estimates BMC which could be used to compare chemical potency across assays. It is possible that production of activity profiles for large sets of chemicals could establish common behavioral phenotypes elucidating common modes of action. This abstract does not necessarily reflect U.S. EPA policy.  

Record Details:

Record Type:DOCUMENT( PRESENTATION/ POSTER)
Product Published Date:05/11/2022
Record Last Revised:07/14/2022
OMB Category:Other
Record ID: 355245