
ALLR

PCC

PCST

FIET

KLD

ED

SW


P value versus sum









1.87 × e03





P value versus AM

3.55 × e^{13}

2.43 × e^{13}

2.35 × e^{13}

6.93 × e^{14}

2.05 × e^{13}

1.38 × e^{12}

2.19 × e^{13}

P value versus GM

7.00 × e^{11}

2.52 × e^{13}

2.04 × e^{12}

2.06 × e^{10}

1.09 × e^{13}

1.58 × e^{12}

1.98 × e^{13}

 The table compares the performance of Tomtom's P values with three other methods for combining column motif comparison scores: summing the raw scores (sum), computing the arithmetic mean (AM), and computing the geometric mean (GM). The comparison is performed for seven different column comparison functions. Each entry in the table is a signed rank P value for the comparison of two ranking methods. '' Indicates that the difference between the two methods is not significant at P = 0.01. All entries correspond to significantly better performance of motif P values than the competing method (Table 1). The table reports results for the S/8 sampling rate. ALLR, average loglikelihood ratio; ED, euclidean distance; FIET, FisherIrwin exact test; KLD, KullbackLeibler divergence; PCC, Pearson correlation coefficient; PCST, Pearson χ^{2} test; SW, SandelinWasserman function.