ABSTRACT
This chapter discusses false discovery rates in genomics data analysis. More reliable statistical methods sometimes come at the price of greater complexity of both mathematics and interpretation of the mathematics. The chapter forms a foundation for understanding and properly using modern methods of interpreting high-dimensional data in biology. It explores a relationship between the local false discovery rate (LFDR) and a nonlocal false discovery rate. Family-wise error rate, a concept that leads to popular adjustments of p values such as the Bonferroni correction. Many estimators of the LFDR have been encoded in free software, most notably in the R statistical computing language (R Development Core Team, 2008). For example, there are LFDR estimators in the R packages locfdr, fdrtool, LFDR.MLE, and PsiHat available from the Comprehensive R Archive Network (CRAN).
