Upper bound for the number of false discoveries among most significant items

Usage

curveMaxFP_BNR2014(p.values, thr)

curveMaxFP_Mein2006(p.values, thr)

Arguments

p.values

A vector containing all $m$ p-values, sorted non-decreasingly

thr

A vector of $K$ JER-controlling thresholds, sorted non-decreasingly

Value

A vector of size $m$ giving an joint upper confidence bound on the number of false discoveries among the $k$ most significant items for all $k \in \{1\ldots m\}$.

Details

These older implementations of 'curveMaxFP' are here for the purpose of testing that the current one yields consistent results.

Author

Gilles Blanchard, Nicolas Enjalbert-Courrech, Pierre Neuvial and Etienne Roquain