Detecting X-Linked Association with a Variance Test of X-Inactivation

The X chromosome (ChrX) plays an important role in animal and human disease. GWAS have identified thousands of loci associated with complex diseases and traits, but most GWAS ignored ChrX due to a few reasons including X-inactivation. In this study, we propose novel tests of X-linked association by capitalizing on X-inactivation. In presence of X-inactivation female heterozygotes are expected to exhibit elevated stochasticity of expressed variants than female homozygotes, which translates to higher variance in the trait for heterozygous females. We propose two types of tests that capitalize on this phenomenon: 1) Test for inflated phenotypic variance; and 2) Weighted regression test. The first type sets out to directly test for higher phenotypic variance of heterozygous than homozygous females. As a test of variance, it is usually less powerful than standard tests of association, which test for mean differences: We conducted a series of simulations, unless the effect size is extremely large, a standard association test had a better performance. However, the variance test tests for overdispersion, hence providing a complementary type of signal. The second type is similar to standard association tests in considering the phenotypic mean, but accounts for the inflation in variance due to X-inactivation. We also conducted simulations to test the performance of the weighted regression. As expected, this test is more powerful than a standard association test in all cases. While our simulations are restricted to a simplistic, these results point to the potential to improve power of detecting X-linked associations when X-inactivation is involved.