We compare the performance of our method to three others, MEME [18], Gibbs Motif Sampler [39], and Projection [22]. We choose these particular methods as they are widely-used, readily accessibly for download via the internet, and can handle the lengths of motifs (11–48 bps) in our dataset. While a recent performance evaluation of motif finders [1] indicates that combinatorial methods such as Weeder [24] have somewhat better performance than other methods, most enumerative combinatorial methods, Weeder included, are not able to handle the lengths of the motifs in our bacterial dataset.