Abstract
Genome wide association studies frequently reveal associations between disease susceptibility and polymorphisms outside coding regions. Such associations cannot always be explained by linkage disequilibrium with changes affecting the transcription products. This has stimulated the interest in characterising sequence variation influencing gene expression levels, in particular in changes acting in cis. Differences in transcription between the two alleles at an autosomal locus can be used to test the association between candidate polymorphisms and the modulation of gene expression in cis. This type of approach requires at least one transcribed polymorphism and one candidate polymorphism. In the past five years, different methods have been proposed to analyse such data. Here we use simulations and real data sets to compare the power of some of these methods. The results show that when it is not possible to determine the phase between the transcribed and potentially cis acting allele there is some advantage in using methods that estimate phased genotype and effect on expression simultaneously. However when the phase can be determined, simple regression models seem preferable because of their simplicity and flexibility. The simulations and the analysis of experimental data suggest that in the majority of situations, methods that assume a lognormal distribution of the allelic expression ratios are both robust to deviations from this assumption and more powerful than alternatives that do not make these assumptions.
Highlights
In recent years, analysis of allelic expression has increasingly been used to ascertain in vivo the influence of sequence variants suspected to affect expression in cis [1,2,3,4]
Statistical Tests: We limit our consideration to previously published approaches or existing methods. These tests assume that the data consist of a set of individuals who have been typed for a putative cis acting polymorphism and transcribed marker, and that allelic expression has been measured in those individuals that are heterozygous for the transcribed marker
First we assess the power of the different tests when allelic expression ratio (AER) follows a log normal distribution (Figure 4), we investigate the case when one transcribed allele is consistently overexpressed and we wish to asses an independent effect of the cis acting polymorphism (Figure 5)
Summary
Analysis of allelic expression has increasingly been used to ascertain in vivo the influence of sequence variants suspected to affect expression in cis [1,2,3,4] Such variants modulate expression from the same chromosome on which they are located include, for example, changes affecting gene promoters or sequence elements regulating message stability. Several recent reports have found that allelic expression analysis can be more powerful in detecting cis acting variants than traditional expression quantitative trait locus (eQTL) analysis [1,2] This is of particular interest since the effects of polymorphisms may vary between tissues and developmental stages [5,6], and assessing these effects in tissues where availability is limited will be facilitated by using more sensitive methods of analysis
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
