Detecting Motifs in a Large Data Set: Applying Probabilistic Insights to Motif Finding