eQTL โ Do Not Interpret Expression Associations as Causal
Why Is This Important?
Disease GWAS identifies associated loci but does not directly reveal which geneโs regulatory processes are altered by the signal. Expression quantitative trait locus (eQTL) analysis narrows down potential regulatory pathways by measuring whether genotype differences and gene expression differences co-vary across individuals.
However, association does not imply that a variant directly causes an expression change. Linkage disequilibrium, cellular composition, and environmental factors can generate similar patterns. eQTLs provide evidence for prioritizing targets for subsequent validation rather than establishing causal conclusions.
Core Concepts
While associations near the gene are typically termed cis-eQTLs and those at distant loci or on different chromosomes trans-eQTLs, distance criteria and analytical definitions must be explicitly specified for each study. sQTLs can be identified by measuring splice junctions or isoforms instead of expression levels; the molecular trait unit selected constitutes part of the result interpretation.
Analyses regress genotype, normalized expression, ancestry, and technical covariates into a regression model to test numerous variantโgene pairs. Although multiple testing correction and hidden factor adjustment are applied, excessive correction may eliminate genuine biological signals. Low sample sizes and allele frequencies hinder the detection of small effects.
Tissue and Cellular Context
Gene regulation varies across tissues, cell types, developmental stages, stimuli, and disease states. eQTLs that are prominent in the liver may not be detectable in blood, and response eQTLs may appear only following immune stimulation. Multi-tissue resources such as GTEx provide a framework for comparing this context-dependent regulation.
Bulk tissue expression represents an average across multiple cell types. When genotype and cell proportions covary, cell composition effects can mimic gene regulation, and signals from rare cell types may be obscured by averaging. Single-cell approaches increase resolution but introduce other limitations, including sparsity, batch effects, and small sample sizes per cell.
When GWAS and eQTL Overlap
Even when the GWAS lead variant and the eQTL lead variant are identical or proximal, it cannot be definitively concluded that a single causal variant underlies both traits. Distinct variants within an LD block may independently influence disease risk and gene expression. Initial assessment should employ conditional analysis and fine-mapping to evaluate independent signals and credible sets.
Colocalization probabilistically compares whether two association patterns are compatible with a shared causal signal. Results depend on assumptions regarding variant coverage, the LD reference panel, the number of causal variants, and prior specifications. A high posterior probability does not definitively prove that the target gene mediates the disease mechanism.
Small Example
Suppose a GWAS locus for an autoimmune disease is associated with low expression of gene A in immune cells. If the two signals colocalize in the relevant cell type, and allele-specific expression and enhancer annotations also support the same direction, gene A becomes a strong candidate. However, causal claims are strengthened only when CRISPR perturbation or independent cohorts reproduce the immune function and disease-related phenotype.
Common Misconceptions
- There is no guarantee that the lead variant is the causal variant.
- The nearest gene is not guaranteed to be the target gene.
- The absence of an eQTL does not imply a lack of regulatory effect, but rather that it was not detected in the measured tissue, condition, or sample.
- Colocalization does not substitute for physical interaction or functional rescue experiments.
Limitations
The generalizability of eQTL resources to other populations and conditions is constrained by the limited ancestry and tissue representation. Expression measurement methods, reference annotations, and pipeline versions also influence results. When using public results, it is essential to document the effect allele, direction, tissue, sample size, and analysis release.
Reading in Context
The process of moving from a GWAS credible set to regulatory mechanisms connects non-coding GWAS and chromatin contact with the concept of chromatin accessibility, which is ATAC-seq.