|
Abstract:
|
Recent publications have highlighted widespread, poor statistical practices in the design, analysis, and reporting of life sciences research. Avoiding such errors is critical to improving translation to the clinic, but applying good statistical practices offers additional rewards - efficient designs, robust results, and sound decisions. In this example, infectious disease scientists worked with a statistician on whole genome siRNA screens against four viruses to identify host proteins that are important to the viral replication cycle. High false positive and false negative rates, however, can make the interpretation of these studies quite challenging, especially when comparing results across multiple viruses. The use of statistical experimental design produced significant improvements in assay quality while investigation of different approaches to the data analysis resulted in a robust analysis workflow, minimizing the false positive and false negative rates across the four screens. This work illustrates the use of good statistical practices to improve the selection and validation of new targets in drug discovery.
|
