On June 3rd, Center Director Andrew Allen gave a talk on methods for analyzing secondary phenotypes in case-control genetic association studies at the SAGES Symposium. This was an opportunity to represent StatGen and Duke’s statistical genetics community at a large, regional conference alongside the discipline’s top researchers.
SAGES brings together an interdisciplinary group of scientists working in the fields of genomics, epidemiology, and statistics. Advances in technology and decreases in the associated costs are driving progress in genomic studies. Studies of whole exome and genome sequences of complex traits in large samples are becoming increasingly common. Other sources of high dimensional information, including expression, epigenetic, metabolic and microbiomic data, are also being collected in disease and control samples. To fully understand the complex bases of human disease and trait variation, all of these factors should be properly considered in a unified analytical framework, together with epidemiological data on environmental exposures and other risk factors.
Allen’s discussion began with a review of case-control studies with two separate samples, cases with the disease and controls without the disease. As measuring exposures is expensive, an oversampling of cases was allowed to minimize the number of exposures that needed to be assessed. While this is an economical approach for assessing associations between genetic exposures and disease, it does not constitute a random sample from the general population. If this isn’t taken into account during the analysis, it can leave to a biased association. Previous studies utilized weighted regression models and also a rare disease model assuming a binary secondary phenotype. Allen’s approach, utilized inverse probability weighted estimating equations from restricted moment model framework which is computationally efficient and, as a result, can be applied genome-wide.
The methodology is further described in a recent article, Robust analysis of secondary phenotypes in case-control genetic association studies, published online on May 30 in Statistics in Medicine. In addition to Allen, authors include lead author, Chuanhua Xing; Janice McCarthy; Josée Dupuis; L. Adrienne Cupples; James B. Meigs; and Xihong Lin. This research is supported by NIH grants: R01DK078616 and R01MH084680.
The symposium was held at the Center for Genetics and Complex Traits (CGACT) of the Perelman School of Medicine at the University of Pennsylvania. The forum provides an opportunity for scientists at all levels in their career to convene and review new developments in these areas of research. The symposium also facilitates exchange of ideas, interactions and collaborations among participants.