Characterizing and Correcting for Selection Bias in High-throughput Problems

Characterizing and Correcting for Selection Bias in High-throughput Problems

Statistics and Genomics Seminar
Jan 28, 2016, 04:00 PM - 05:00 PM | 1011 Evans Hall | Happening As Scheduled
Noah Simon, Department of Biostatistics, University of Washington
There is a growing scientific interest in simultaneously estimating many effect-sizes (or means) in high-throughput experiments. Examples include differential expression experiments, genome wide association studies, and interaction screens, among others. Typically we are most interested in those effect-sizes, which look most extreme in our sample. Unfortunately by selecting those "most extreme"...