We calculated the scores of the predictions of Weeder and ANN-Spec and the planted motifs. The planted motif has a higher score than the predictions of the tools for most datasets, as illustrated in Figure 4. The obvious gap between the scores of planted binding sites and the predictions reflects a lack of optimum of the search strategies adopted by these tools. Recall that ANN-Spec is a generalized version of SEM (Stochastic EM), and Weeder uses a greedy and heuristic search method.