Seminars

Statistical Science Departmental Seminar - March 1

Testing for Association in Genome-wide Association Studies

Speaker :   Dr. Zhongxue Chen, Assistant Professor,   Epidemiology and Biostatistics Department, Indiana University Bloomington   

Abstract:

In genetic association studies, due to the varying underlying genetic models, there exists no single statistical test that is most powerful under all situations. Current studies show that if the underlying genetic models are known, trend-based tests, which outperform the classical Pearson's chi-square test, can be constructed. However, when the underlying genetic models are unknown, chi-square test is usually more robust than trend-based tests. We propose a generalized genetic model, order restricted relative risk (O3R), which includes many special ones, such as dominant, recessive, additive, and log additive models.  A new statistical testing procedure is then derived for the O3R model. Through a Monte Carlo simulation study, we show that this new method is generally more powerful than the chi-square test, and more robust than trend test. The proposed method is illustrated using real datasets on breast cancer and prostate cancer.