Efficient statistical method for association analysis of x-linked variants

Abstract

Unlike the gene-poor Y-chromosome, the X-chromosome contains over 1,000 genes that are essential for proper development and cell viability. Thus, amount of females X-linked gene expression would be expected to be twice as males. To adjust this unbalancedness, one of two X-linked genes in female is often inactivated and this is known as X-chromosome inactivation (XCI). However, recent studies found that a gene for inactivation can be nonrandomly selected from two X-linked genes and XCI was not observed in some X-linked genes. This biologically complicated process has prevented the efficient statistical association analyses, and it may partially explain the relatively small finding of significantly associated X-linked variants. Here, I propose a new statistical method robust against the uncertain biological process. The proposed method consists of two steps. First, p-values for various biological processes are calculated and combined into a single p-value with modified Fishers method and minimum P-value. Our simulation results show that the scenario for specific XCI process with true underlying coding is the most powerful but the proposed methods are the second best.