Abstract:
|
Allele-specific expression (ASE) can be quantified by the relative expression of two alleles in a diploid individual, and such expression imbalance may explain phenotypic variation and disease pathophysiology. Existing methods for gene-based ASE detection can only analyze one individual at a time, thus wasting shared information across individuals. To overcome this limitation, we develop GLMM-seq, a generalized linear mixed-effects model that can simultaneously model multi-SNP and multi-individual information. The model is able to detect gene-level ASE under one condition and differential ASE between two conditions (e.g., diseased vs. healthy controls). To model multiple individuals simultaneously, we further extend existing individual-based ASE detection methods using a weighted ordered p-value approach . Extensive simulations indicate that our methods perform consistently well under a variety of scenarios. We further apply our methods to real data in the Genetics of Evoked Response to Niacin and Endotoxemia Study, and our results will provide novel candidates for modulation of innate immune responses in humans.
|