Comparisons between FMotif and CisFinder on large (, ) samples with , , and no () noise sequences.

‘Planted Motif’ indicates a planted motif consensus in a set of sequences, ‘FMotif (Top-1)’ indicates the top ranked motif consensus found by FMotif in a sample set, ‘CisFinder’ indicates the closest matching motif consensus (described by IUPAC nucleotide codes) found by CisFinder in a sample set, ‘’ indicates the number of variants of a reported motif found by FMotif or CisFinder in a sample set, and ‘Rank’ after ‘’ is the ranking number of the reported motif found by Cisfinder. The site-level sensitivity () and positive predictive value () metrics described by Tompa [24] were used to statistically quantify the accuracy of the two methods, where and , is the number of known sites overlapping predicted sites, is the number of known sites not overlapping predicted sites, and is the number of predicted sites not overlapping known sites. A predicted site overlaps a known site if they share at least a half of the length of known sites.