Figure 5
From: A cross-validation scheme for machine learning algorithms in shotgun proteomics
Percolator's algorithm for normalizing the scores from different cross-validation sets. The algorithm takes two inputs: a set containing sets of PSMs, and a significance threshold α. Each PSM is represented as a tuple: a score and an accompanying boolean indicating whether this is a decoy PSM. The function qValue takes as input a set of scored PSMs and finds the minimal score that achieves the specified significance α, and MedianDecoy returns the median decoy score from the given set. The function returns a combined collection of normalized scores.