Compound Ranking in High-Throughput Screening by Using Time Course Data from Cell-Based Assays

• Activity Number: 613
• Type: Contributed
• Date/Time: Thursday, August 2, 2012 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistical Learning and Data Mining
• Author(s): Francisco Diaz*+
• Companies: University of Kansas Medical Center
• Keywords: High throughput screening ; anticancer drugs ; exponential growth model ; drug effectiveness ; cytotoxicity ; IC50

Abstract

The half maximal inhibitory concentration (IC50) has several limitations that make it unsuitable for examining a large number of compounds in cytotoxicity studies, particularly when multiple exposure periods are tested. We describe a new mathematical measure of drug effectiveness, which allows ranking compounds according to their toxic effects on live cells. This measure, which can be easily implemented in linear regression software, combines all exposure times tested and compares the growth rates of a particular cell line in the presence of the compound with its growth rate in the presence of DMSO alone. Our approach allows measuring a wider spectrum of toxicity than the IC50 approach, and allows automatic analyses of a large number of compounds. The approach provides a comparable measure of effectiveness for each investigated compound (both toxic and non-toxic), and allows statistically testing the null hypothesis that a compound is non-toxic versus the alternative that it is toxic. As an illustration, we describe the results of a cell-based study of the cytotoxicity of 24 analogs of novobiocin, a C-terminal inhibitor of heat shock protein 90 (Hsp90).