Addressing Unusual Assay Variability with Robust Statistics. (December 2021)

• Record Type: Journal Article
• Authors: Haelewyn, Jason; Iversen, Philip W.; Weidner, Jeffrey R.
• Is Part Of: SLAS discovery. Volume 26:Number 10(2021)
• Page Start: 1291
• Page End: 1297
• Publication Date: 2021-12
• Subjects: robust statistics -- data analysis -- bioassay; Drugs -- Analysis -- Periodicals; Drugs -- Testing -- Periodicals; Biomolecules -- Analysis -- Periodicals; Biomolecules -- Analysis; Drugs -- Analysis; Drugs -- Testing; Drug Evaluation, Preclinical; Molecular Biology -- methods; Periodicals; Periodicals; 615.1
• DOI: 10.1177/24725552211038379
• Languages: English
• ISSNs: 2472-5552

Abstract

Well-behaved, in vitro bioassays generally produce normally distributed values in their primary (efficacy) data. Accordingly, the best practices for statistical analysis are well documented and understood. However, assays may occasionally display unusually high variability and fall outside the assumptions inherent in these standard analyses. These assays may still be in the optimization phase, in which the source of variation could be identified and addressed. They might also represent the best available option to address the biological process being examined. In these cases, the use of robust statistical methods may provide a more appropriate set of tools for both data analysis and assay optimization. This article provides guidance on best practices for the use of robust statistical methods for the analysis of bioassay data as an alternative to standard methods. Impacts on experimental design and interpretation will be discussed.