Testing Strategies for Genome-Wide Association Studies: Using the Same Dataset for Screening and Replication

• Activity Number: 39
• Type: Invited
• Date/Time: Sunday, July 29, 2007 : 4:00 PM to 5:50 PM
• Sponsor: ENAR
• Author(s): Christoph Lange*+
• Companies: Harvard School of Public Health
• Keywords: Family-based studies ; Genome-wide association ; Multiple testing problem ; Replication ; Screening ; Two-stage

Abstract

Genome-wide association study (GWAS) is now a reality. Their success will depend on the efficient handling of the statistical challenges posed by GWAS. GWAS must translate the markedly increased amount of SNP-information into increased statistical power. For the number of statistical tests computed in a GWAS, standard statistical methods for handling the multiple testing problem (MTP) are too conservative. For QTL, VanSteen et al (2005) proposed a new approach that bypasses the MTP in family-based studies. The approach consists of a screening step and a testing step, which handle the MTP all in the same data set. It outperforms standard methodology in terms of statistical power (e.g., it addresses the MTP and the replication problem both at the same time). We will review the VanSteen approach, and discuss extensions to dichotomous traits in trio designs and to case/control design.