Recent advances in molecular technologies have resulted in the ability to screen hundreds of thousands of single nucleotide polymorphisms and tens of thousands of gene expression profiles. While these data have the potential to inform investigations into disease etiologies and advance medicine, the question of how to adequately control both type I and type II error rates remains. Genetic Analysis Workshop 15 datasets provided a unique opportunity for participants to evaluate multiple testing strategies applicable to microarray and single nucleotide polymorphism data. The Genetic Analysis Workshop 15 multiple testing and false discovery rate group (Group 15) investigated three general categories for multiple testing corrections, which are summarized in this review: statistical independence, error rate adjustment, and data reduction. We show that while each approach may have certain advantages, adequate error control is largely dependent upon the question under consideration and often requires the use of multiple analytic strategies.
(c) 2007 Wiley-Liss, Inc.