Figure 5

Results of Wilcoxon tests with the alternative hypothesis thatthe unpaired probabilities in the predicted binding sites are higher than negative controls randomly drawn from (A) the same positive gold-standard sequences, and (B) random sequences.