The power of independent types of genetic information to detect association in a case-control study design

Authors

  • Sungho Won,

    1. Department of Epidemiology and Biostatistics, Case Western Reserve University, Cleveland, Ohio
    Search for more papers by this author
  • Robert C. Elston

    Corresponding author
    1. Department of Epidemiology and Biostatistics, Case Western Reserve University, Cleveland, Ohio
    • Department of Epidemiology and Biostatistics, Case Western Reserve University School of Medicine, Wolstein Research Building, 10900 Euclid Avenue, Cleveland, OH 44106-7281
    Search for more papers by this author

Abstract

There have been many single nucleotide polymorphism-based tests suggested for association analysis in a case-control design. The possible evidence for association comprises three types of information: differences between cases and controls in allele frequencies, in parameters for Hardy Weinberg disequilibrium (HWD) and in parameters for linkage disequilibrium (LD). Here, first we find the pairwise covariances between statistics that measure these three types of information and show that the statistics are asymptotically trivariate normally distributed. Then we compare their power analytically to determine the most informative statistics according to the disease model. Our results show that differences in parameters for HWD are informative for dominant and recessive disease models, while differences in allele frequencies and in parameters for LD are generally informative except for rare recessive disease models. There is mutual independence of the statistics that detect these three differences under Hardy Weinberg equilibrium at the marker locus and linkage equilibrium between markers in the population. Knowing the pairwise covariances between the statistics makes it possible to define statistics that are mutually independent. This allows us to perform sequential analyses of the same data without the need to adjust significance levels for all the multiple analyses being performed on the same data set. As a result we can have improved flexible strategies to increase the power of genome-wide association studies without requiring the collection of a new, independent sample. Genet. Epidemiol. 2008. © 2008 Wiley-Liss, Inc.

Ancillary