Efficiency Robust Tests for Linkage or Association Joseph L. Gastwirth
George Washington University and the Division of Cancer Epidemiology and Genetics, National Cancer Institute
When testing whether a candidate gene is associated with a disease or whether a marker is in linkage with a disease-causing gene, often the mode of inheritance is not known. This creates a problem as the optimal test statistic depends on the genetic model underlying the disease, e.g., the means test used to analyze sib-pair data is optimum for an additive mode of inheritance in which each disease-causing allele increases the disease-risk or penetrance by the same amount. For the recessive model, the proportions test is the most powerful one. The theory of efficiency robustness is used to determine a compromise test that has high power relative to the optimal test for each of the scientifically plausible model. After reviewing the concepts of efficiency robustness, we present efficiency robust tests for sib-pair linkage studies, TDT-type candidate gene association studies and case-control association studies which have high power for a wide family of scientifically plausible genetic models. The talk is based on joint work with Dr. Boris Freidlin, Dr. Marvin Podgor, Dr. Zhaohai Li and Dr. Gang Zheng.