Rcurvep: Concentration-Response Data Analysis using Curvep

Provide an R interface for processing concentration-response datasets using Curvep, a response noise filtering algorithm. The algorithm was described in the publications (see references below). Other parametric fitting approaches (e.g., Hill equation) are also adopted for ease of comparison. Also, methods for calculating the confidence interval around the activity metrics are also provided. The methods are based on the bootstrap approach to simulate the datasets. The simulated datasets can be used to derive the baseline noise threshold in an assay endpoint. This threshold is critical in the toxicological studies to derive the point-of-departure (POD).

Details

Different strategies are used to simulate the datasets:

  • Curvep - bootstrapping the responses of replicates at each concentration

  • Hill equation - bootstrapping the residuals and adding back to the fitted responses (by Hill) at each concentration

For Curvep the bootstrapping strategy is different depending on the type of datasets. Datasets can be grouped into three types:

  1. dichotomous binary incidence data (e.g. mortality data from alternative animal model data)

  2. continuous data with high number of replicates (e.g. alternative animal model data)

  3. continuous data with low number of replicates (e.g. in vitro data)

Bootstrapping strategies:

  1. bootstrap incidence out of total animals per concentration then calculate percentage of incidence

  2. bootstrap replicate responses per concentration directly

  3. bootstrap vehicle control responses and add back to the fitted responses by linear regression per concentration (experimental)

To learn more about Rcurvep start with the vignettes: browseVignettes(package = "Rcurvep")

References

Curvep

Sedykh A, Zhu H, Tang H, Zhang L, Richard A, Rusyn I, Tropsha A (2011-March). “Use of in vitro HTS-derived concentration-response data as biological descriptors improves the accuracy of QSAR models of in vivo toxicity.” Environmental health perspectives, 119, 364--370. doi: 10.1289/ehp.1002476 , http://europepmc.org/articles/PMC3060000.

Sedykh A (2016). “CurveP Method for Rendering High-Throughput Screening Dose-Response Data into Digital Fingerprints.” Methods in molecular biology (Clifton, N.J.), 1473, 135--141. doi: 10.1007/978-1-4939-6346-1_14 .

Bootstrap

Hubbard TD, Hsieh J, Rider CV, Sipes NS, Sedykh A, Collins BJ, Auerbach SS, Xia M, Huang R, Walker NJ, DeVito MJ (2019-March). “Using Tox21 High-Throughput Screening Assays for the Evaluation of Botanical and Dietary Supplements.” Applied in vitro toxicology, 5, 10--25. doi: 10.1089/aivt.2018.0020 , http://europepmc.org/articles/PMC6442399.

Hsieh J, Ryan K, Sedykh A, Lin J, Shapiro AJ, Parham F, Behl M (2019-January). “Application of Benchmark Concentration (BMC) Analysis on Zebrafish Data: A New Perspective for Quantifying Toxicity in Alternative Animal Models.” Toxicological sciences an official journal of the Society of Toxicology, 167, 92--104. doi: 10.1093/toxsci/kfy258 , http://europepmc.org/articles/PMC6317423.