Since thousands of molecular phenotypes are tested genome-wide, an FDR correction is commonly applied. This estimates the proportion of false positive findings, known as the FDR, by comparing the number of hits declared to the number that would be expected by chance. The Benjamini–Hochberg (BH) procedure (Benjamini and Hochberg, 1995) is one way of controlling the number of false positive results. However, this is too conservative in most of the QTL studies where we expect a substantial fraction of the phenotypes to be affected by genetic variants. To account for this, it is recommended instead the use of the FDR procedure described by Storey and Tibshirani (ST) (Storey and Tibshirani, 2003) which fits particularly well in this context. The ST procedure assumes that the set of association tests originates from a mixture of both the null and the alternative hypothesis and estimates π0, defined as the proportion of hypotheses for which the null is true. Implicitly, the BH procedure assumes π0 is 1, whereas the ST procedure learns it from the data, resulting in more statistically significant hits. Of note,
