pi1.R

library(qvalue)

p = runif(1000, 0, 1)

# compute pi0, the fraction of tests where the null is supported
# so pi1 = 1-p0 is the fraction of tests where the null is rejected
# This method doesn't depend on p-value cutoffs and works even of nothing
# is genome-wide significant
pi1 = 1 - qvalue(p)$pi0

# Here pi is near zero because the p-values are uniform
pi1