A0372
Title: Meta-analysis in family-based study of disease subtypes
Authors: Debashree Ray - Johns Hopkins University (United States) [presenting]
Abstract: Family-based designs often examine genetic determinants of child health outcomes and rare diseases (e.g., case-parent trio design consisting of an affected child and both parents). When investigating similarities and differences in the genetic basis of subtypes of such diseases, investigators have typically used meta-analysis techniques. However, a meta-analysis in this context tests the global null hypothesis that a genetic marker does not affect any disease. This is not exactly the null hypothesis to test when the goal is to identify a common genetic basis. A new statistical approach is discussed for detecting the genetic overlap of two diseases or disease subtypes by considering a composite null hypothesis that a genetic marker is associated with none or only one of the traits. A mixture distribution is used for the null distribution of the test statistic that allows for fractions of millions of genetic markers to be associated with none or only one of the traits. An asymptotic approximation of the null distribution IS useD that avoids estimating nuisance parameters related to mixture proportions and variance components. Our method requires only summary-level data as used in the meta-analysis, has well-calibrated type I error at stringent levels used in genetic studies, and can achieve major power gain over alternative methods typically used in the literature. Finally, an application is shown to the case-parent trio study of orofacial cleft subtypes.
