A general framework for two-stage analysis of genome-wide association studies and its application to case-control studies.
Wason, James MS;
Dudbridge, Frank;
(2012)
A general framework for two-stage analysis of genome-wide association studies and its application to case-control studies.
American journal of human genetics, 90 (5).
pp. 760-773.
ISSN 0002-9297
DOI: https://doi.org/10.1016/j.ajhg.2012.03.007
Permanent Identifier
Use this Digital Object Identifier when citing or linking to this resource.
Two-stage analyses of genome-wide association studies have been proposed as a means to improving power for designs including family-based association and gene-environment interaction testing. In these analyses, all markers are first screened via a statistic that may not be robust to an underlying assumption, and the markers thus selected are then analyzed in a second stage with a test that is independent from the first stage and is robust to the assumption in question. We give a general formulation of two-stage designs and show how one can use this formulation both to derive existing methods and to improve upon them, opening up a range of possible further applications. We show how using simple regression models in conjunction with external data such as average trait values can improve the power of genome-wide association studies. We focus on case-control studies and show how it is possible to use allele frequencies derived from an external reference to derive a powerful two-stage analysis. An illustration involving the Wellcome Trust Case-Control Consortium data shows several genome-wide-significant associations, subsequently validated, that were not significant in the standard analysis. We give some analytic properties of the methods and discuss some underlying principles.