Table 1.

Overview of methods. The match model is the consensus representation of a single motif, motif combination is how the component scores of a composite motif are combined, and distance score is how the conservation of inter-motif distances within a composite motif is modeled.

ALGORITHM NAME	MATCH MODEL	MOTIF COMBINATION	DISTANCE SCORE
Weeder [42]	mismatch	-	-
Dyad analysis [35]	oligos	dyad1	constraint
MCAST [71]	PWM	sum	gap penalty
REDUCE [67]	PWM	dyad	constraint2
MDScan [87]	PWM	-	-
Gibbs sampler [97]	PWM	intersection3	uniform
MEME [98]	PWM	-	-
LOGOS [73]	DM	HMM	distribution
Motif regressor [89]	PWM	-	-
ModuleSearcher [70]	PWM	sum	window4
Stubb [48]	PWM	HMM	window
GANN [60]	flexible	ANN5	window
ANN-Spec [86]	PWM	-	-
(Wasserman) [58]	PWM	Logistic regr.	window
CoBind [68]	PWM	sum	window
Cister [72]	PWM	HMM	distribution
SeSiMCMC [122]	PWM	-	-
SMILE [40, 123]	mismatch	intersection	constraint
BioProspector [49]	PWM	sum	constraint
(Segal) [94]	PWM	-	-
(Sinha) [33]	reg.exp	dyad	constraint
ConsecID [56]	PWM	intersection	window
SCORE [69]	IUPAC	intersection	window
Gibbs recursive [52]	PWM	mixture model	distribution
(Hong) [95]	PWM	-	-
AlignACE [124]	PWM	-	-
Improbizer [117]	PWM	-	-
CisModule [119]	PWM	mixture model	mixture model
(Thompson) [66]	PWM	Markov model	constraint

¹Two single motifs that both have to occur

²Separate constraints on each inter-motif distance

³Several single motifs that all have to occur

⁴All single motifs have to occur within a sequence window of restricted length

⁵Artificial neural network