far, a commonly employed permutation strategy relies on performing a fixed number of permutations per phenotype (1000–10 000) to control the running times at the cost of accurately assessing the statistical significance of the most strongly associated QTLs. Here, we present FastQTL, a user- and cluster-friendly QTL mapper, which improves upon Matrix eQTL by implementing a fast and efficient permutation scheme in which the null distribution of associations for a phenotype is modeled using a beta distribution. This allows us to approximate the tail of the null distribution relatively well using only few permutations, and then to accurately estimate adjusted P-values at any significance level in short running times.
