Table 1 Performance of the hill climbing algorithm on well-characterized S. cerevisiae regulons. The reported binding sites were each run through the hill climbing algorithm to determine the algorithm's ability to generate a representative consensus motif given only one instantiation. Abbreviations are as follows: |B|-the number of genes in the regulon; HC(b)-the consensus motif reported by HC(·) that had the highest Φ score; ΔSig-the maximum change in Sig score due to HC for any binding site; Φ b-the maximum Φ score for any one binding site; Φ HC-the Φ score after HC is run on the known binding sites.

13

gagtca

17.80

0.36

0.36

3

cwcgtgrm

17.26

0.61

0.55

6

tcmwttcayccg

28.33

0.27

0.29

5

gatwas

40.66

0.73

0.56

20

bagtcab

26.10

0.29

0.44

9

ctyhc

12.59

0.43

0.18

3

ytaatctaatc

10.28

0.63

0.64

5

gttggttggtgga

13.37

0.37

0.23

3

sactaattaggaaa

10.00

0.33

0.36

12

brdgtaadt

29.17

0.21

0.48

43

chnwttmggdaa

53.31

0.05

0.29

4

wcgcg

40.46

0.62

0.78

13

ccccrbwww

19.43

0.21

0.38

10

hyccrcggr

138.96

0.50

0.71

6

gtaaattagttaatt

0.00

0.40

0.21

12

cacgtggracta

8.41

0.13

0.14

18

acacccagacmkc

12.98

0.09

0.15

20

bvywacccs

25.01

0.36

0.47

8

ycyattgttctc

14.86

0.13

0.38

3

tctcggcggtta

0.00

0.34

0.34

8

cdcgawa

37.79

0.58

0.72

4

aggtmaacaa

5.85

0.25

0.50

6

atgmaac

24.74

0.48

0.53

20

yatava

18.44

0.44

0.20

4

ttgccmttmgc

16.63

0.25

0.40

21

gwgtagtgaca

12.12

0.05

0.05

23

cgtggatgaaac

6.44

0.04

0.04

5

aaaatagcgcctc

0.00

0.22

0.20

12

wdwtwgccsccvw

61.10

0.15

0.65

11

24.21

0.33

0.39