Z’ Does Not Need to Be > 0.5
Haim Bar, Adam Zweifach
Abstract
The assay metric Z' has come to play a critical gatekeeping role in determining whether high-throughput assays can be performed. While Z' is commonly required to be > 0.5, this expectation is not well supported. Requiring Z' > 0.5 likely prevents many potentially useful phenotypic and cell-based screens from being conducted, and causes other assays to be conducted under extreme conditions that may prevent activity from being found. We used power analysis and a novel numerical simulation approach to determine how Z' reflects assay performance under a variety of conditions. Our results show that assays with Z' > 0.5 perform better than assays with lower Z', but when an appropriate threshold is selected, assays with Z' < 0.5 can almost always find useful compounds without generating too many false positives. We provide a method that will allow researchers to estimate how to set an appropriate threshold for their assay. We suggest that instead of always requiring Z' > 0.5, assays with Z' < 0.5 should be performed when they can be justified in terms of the importance of the target and the limitations of alternate assay formats.