An eQTL is a locus that explains a fraction of the genetic variance of a gene expression phenotype. I.e. the difference in expression is between individuals.
An eQTL study is basically a GWAS but using expression as the trait (instead of e.g. T1D).
The same resolution issue occurs as in GWAS and for this reason cis-eQTLs (e.g. in promoters) are easier to detect as you know that they are proximal to the gene/loci (have fewer possibilities to investigate). Trans eQTLs require way more tests (multiple testing problem - need way more power).
ASE occurs when expression levels vary at the alleles of a heterozygous individual. I.e. the difference in expression of the alleles is a within individual comparison.
ASE can be used to find cis-eQTLs.
Consider a mutation in the promoter region of a gene at one allele, this mutation may cause a transcription factor to preferentially bind over the wild-type allele and therefore the mutated allele will be more highly expressed. This means that across the population, individuals that carry this altered allele will have higher expression of this gene (with transcripts expressing that altered allele) than individuals that carry the wild-type allele 1. I.e. the two alleles are expressed at different rates.
eQTLs are cell type and cell state (e.g. environmental factors) specific. So that, the cell type in which the eQTL effect is observed often matches cell types already thought to be relevant to the disease in question. In other words, disease phenotypes may only be visible in certain tissues.
Should move away from averaging results over individual cell types (and even whole tissues - e.g. whole blood) to measuring them in individual cells. One study did this and found that many eQTLs are only detectable when studying single cell, and these would have been missed when the expression is averaged over multiple cells. http://dx.doi.org/10.1038/nbt.2642.
Research implies that CVs are enriched in cis regulatory elements (CREs) that are active in disease-relevant cell types and thus that CVs influence disease risk by altering the function of cell type-specific regulatory elements (supported by the overlap of CVs with eQTLs).
Colocalisation: Two independent association signals at the same loci (e.g. one from eQTL analysis and one from disease-association) are consistent with a shared CV.
Investigate whether two independent association signals at the same loci are consistent with a shared CV –> “colocalised traits” which are more likely to share a causal mechanism. For example, can be used to integrate disease scans (e.g. case-control GWAS) with eQTL studies.
This integration of results may help to identify the causal gene and the tissue in which the effect is mediated (since eQTLs tend to be tissue dependent).
Previous colocalisation methods assume that if there are multiple CVs in a region, then all (or none) are shared.
Coloc assumes that the CV has been directly typed or well imputed, and that at most one association is present for each trait in the genomic region of interest (e.g. the regions have reached genome-wide significance).
Coloc calculates the posterior probability of each hypothesis (\(H_0\): No association with either trait, \(H_1(/2)\): Association with only trait 1(/2), \(H_3\): Associations with both traits but at independent SNPs, \(H_4\): Associations at both traits, at a shared SNP), where \(H_4\) is the case of colocalisation. The method uses ABFs and does not require iterative computation schemes (e.g. MCMC).
Note that the aim of coloc is different from fine-mapping; it is not trying to pinpoint the CV but only whether a shared CV is plausible.
Assumes each SNP a priori equally likely to be causal, but perhaps this should vary depending on things like distance to gene or functional elements in this chromosome region.
Note that GWAS signals can be explained by eQTLs only when the CV affects the phenotype by altering the amount of mRNA produced, but not when the phenotype is affected by changing the type of protein produced (e.g. missense mutation).