False positive rate inflation and admixed populations in the quantitative transmission disequilibrium test
Wednesday, September 14, 2016 — Poster Session I
- AM Musolf
- JE Bailey-Wilson
The transmission disequilibrium test (TDT) is a well-known statistical test for the analysis of trio data that functions as both a test of linkage and a test of association and is robust to the problem of population stratification. TDT has been extended to quantitative phenotypes and this study aims to examine the robustness of the quantitative TDT when phenotypic means differ across populations. Genome-wide trio data was simulated for two distinct populations using HapMap allele frequencies. Phenotype data was simulated from a normal distribution and assigned to individuals at random. In the equal-mean scenario, data was simulated for both populations using the same mean. In the different-mean scenario, data for each population was drawn same distribution but using different means. Type I error rates were calculated after the populations were analyzed by qTDT both alone and mixed. We have found that the qTDT statistic maintains proper type I error when the phenotypic means are equal, but results in inflation if the means differ across populations. Larger differences between the means of the populations lead to higher levels of type I inflation. This is a critical deficiency in this test, since inflation of false positive rates in the presence of population admixture is the main reason that TDT tests are utilized. We have developed several corrections that restore proper type I error levels. Analyses to determine how these corrections affect power are ongoing.
Category: Genetics and Genomics