We evaluate the overlap between motif predictions made by our approach and the known motifs using the nucleotide level performance coefficient (nPC) [1,17]. Let nTP, nFP, nTN, nFN refer to nucleotide level true positives, false positives, true negatives and false negatives respectively. For example, nTP is the number of nucleotides in common between the known and predicted motifs. The nPC is defined as nTP/(nTP + nFN + nFP); it is a stringent statistic, penalizing a method for both failing to identify any nucleotide belonging to the motif as well as falsely predicting any nucleotide not belonging to the motif. Though nPC takes both false positives and false negatives at the nucleotide level into account, we also find it useful to consider site level statistics. Following [1], we consider two sites to be overlapping if they overlap by at least one-quarter the length of the site. Defining site level statistics similarly to the nucleotide level statistics above (e.g., site level true positives, sTP, is the number of known sites overlapped by predicted sites), site level sensitivity sSn is sTP/(sTP + sFN).