. 2015 Apr 16;31(16):2601–2606. doi: 10.1093/bioinformatics/btv201

Table 1.

Formulae for P-value calculation

Inputs	A SNP set of LD-pruned r GWAS-index SNPs
	Regulatory regions of interest formatted as BED files
	m = number of control SNPs selected for each index SNP

Intermediate Statistics	SNP set i (1 ≤ i ≤ r) = index SNP i and its m control SNPs
	$p_{i} = \frac{number of SNPs in SNP set i fall in regulatory regions of interest}{m + 1}$
	$S_{i} = {\begin{matrix} 1 & randomly drawn SNP from SNP set i falls in regulatory regions of interest \\ 0 & otherwise \end{matrix}$
	S_i ∼ Bernoulli (p_i)
	$\sum_{i = 1}^{r} S_{i} \sim sum of r independent non–identical Bernoulli distribution$

Outputs	s = number of SNPs that fall in regulatory regions of interest in the input GWAS-index SNPs
Outputs	Enrichment P-value = P $(\sum_{i = 1}^{r} S_{i} \leq s)$
	Expected value of $\sum_{i = 1}^{r} S_{i}$

A SNP is considered to be falling in regulatory regions of interest if itself or any of its LD proxies has positional overlap with the regions.