You've Gotta Be Lucky: Coverage and the Elusive Gene–Gene Interaction

Authors

  • Matthew Reimherr,

    1. Department of Statistics, The University of Chicago, 5734 S. University Ave., Chicago, IL
    Search for more papers by this author
  • Dan L. Nicolae

    Corresponding author
    1. Department of Statistics, The University of Chicago, 5734 S. University Ave., Chicago, IL
    2. Department of Medicine, The University of Chicago, 900 East 57th Street, Chicago, IL
      Corresponding author: Dan L. Nicolae, Departments of Statistics and Medicine, The University of Chicago, 5734 S. University Ave., Chicago, Illinois 60637. Tel.: 773-702-4837; Fax: 773-702-9810; E-mail: nicolae@galton.uchicago.edu
    Search for more papers by this author

Corresponding author: Dan L. Nicolae, Departments of Statistics and Medicine, The University of Chicago, 5734 S. University Ave., Chicago, Illinois 60637. Tel.: 773-702-4837; Fax: 773-702-9810; E-mail: nicolae@galton.uchicago.edu

Summary

Genome-wide association studies (GWAS) have led to a large number of single-SNP association findings, but there has been, so far, no investigation resulting in the discovery of a replicable gene–gene interaction. In this paper, we examine some of the possible explanations for the lack of findings, and argue that coverage of causal variation not only has a large effect on the loss in power, but that the effect is larger than in the single-SNP analyses. We show that the product of linkage disequilibrium measures, r2, between causal and tested SNPs offers a good approximation to the loss in efficiency as defined by the ratio of sample sizes that lead to similar power. We also demonstrate that, in addition to the huge search space, the loss in power due to coverage when using commercially available platforms makes the search for gene–gene interactions daunting.

Ancillary