eQTL Analysis — Expression Quantitative Trait Loci Analysis

A cis-eQTL is a variant whose effect on gene expression is detected within a defined genomic window around the gene it regulates (commonly ±1 Mb). A trans-eQTL lies outside that window and acts at a distance — often through transcription factors or other intermediaries. Cis-eQTLs are more numerous, easier to detect, and less prone to false positives; trans-eQTLs require very large samples and strict multiple-testing thresholds because the search space is vastly larger.

For cis-eQTL detection of moderate-effect variants, n ≈ 100–200 is the minimum often cited, but power increases substantially up to n ≈ 500 and beyond. Robust trans-eQTL mapping typically requires n > 1000. The required sample size also depends on effect size, allele frequency, expression noise, and the stringency of the multiple-testing correction. Power calculators such as those provided in the FastQTL documentation can inform study design.

Colocalisation analysis — for example, using the coloc R package or its extensions (HyPrColoc for multi-trait) — formally tests whether an eQTL and a GWAS signal at the same locus are driven by the same causal variant or merely sit in the same region by coincidence. A posterior probability of colocalisation (PP4) above ~0.8 is conventionally taken as evidence that the two signals share a causal variant.

Yes, but single-cell eQTL mapping (sc-eQTL) requires additional methodological considerations: aggregating counts to pseudo-bulk profiles per donor per cell type before testing, accounting for variable numbers of cells per donor, and correcting for cell-type-proportion differences. Tools such as tensorQTL with pseudo-bulk input or dedicated sc-eQTL frameworks (e.g., scPower, FactorQTL) are recommended over naive application of bulk-tissue pipelines.

Matrix eQTL (R package) is widely used for its speed and flexibility. FastQTL and TensorQTL (GPU-accelerated) are standard choices for large cohorts. PEER or similar methods handle latent factor covariate estimation. For colocalisation, the coloc R package is the community standard. The GTEx analysis pipeline is publicly documented and serves as a reproducible reference implementation.