Power and Sample Size Estimation for Genome-Wide Association Studies

• Activity Number: 550
• Type: Contributed
• Date/Time: Wednesday, August 7, 2013 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Wei-Jiun Lin*+ and James Chen and Kuang-Fu Cheng
• Companies: Feng Chia University and National Center for Toxicological Research, FDA and Graduate Institute of Biostatistics and Biostatistics Center, China Medical University
• Keywords: genome-wide association studies ; power and sample size analysis ; linkage disequilibrium ; multiple risk alleles

Abstract

The statistical power for genome-wide association studies is commonly defined as the (weighted) average power over all SNPs in the reference panel or the probability of detecting at least one of susceptibility loci for complex diseases/traits. The estimated sample size based on these power formulas may lead to a less reliable study because the probability of detecting multiple risk alleles is unknown. Here, the power is formulated as the probability of identifying at least a specified number of risk alleles. The probability depends on the distribution of the number of true positives. We propose a method to estimate the distribution which takes linkage disequilibrium between risk alleles into account. The simulation study shows that the sample size calculated based on the proposed method ensures identifying at least the specified number of risk alleles with high probability, whereas the needed sample size based on the common power formulas may result in the probability lower than 50%.