X chromosome inactivation across primary human tissues is mostly complete, with significant implications for genetic and clinical studies

Abstract Background X chromosome inactivation (XCI) refers to silencing of genes on one copy of the X chromosome in XX females, resulting in dosage compensation between XX females and XY males. Genes can escape this silencing, potentially leading to sex-based differences in disease. Results Across three primary tissue types, we determined XCI status by integrating whole-genome sequencing with bulk RNA-Seq data to assess allele-specific expression (ASE) at heterozygous SNPs. Across all genes and tissues, the average percentage of individuals showing escape was 4.7%. We show that models of full dosage compensation and no dosage compensation are strongly correlated over most parameter space. For G6PD deficiency, we illustrate that in G6PD*B/G6PD*A- heterozygotes, even if silencing is complete and escape is not associated with disease risk, the allele that is expressed can affect mRNA abundance. Conclusions With respect to studies of the genetic architecture of complex traits, our results suggest that a model of full dosage compensation, although not strictly correct for much of chromosome X, is more appropriate than a model of no dosage compensation. We conclude that uncertainty about the model of dosage compensation should not be an impediment to analysis of chromosome X in genetic epidemiology studies.