An integrative computational approach to effectively guide experimental identification of regulatory elements in promoters

Table 4 Coverage values of modules on positive, negative and random datasets

Regulatory module	Positive promoter set		Negative promoter set		Random genomic set
Regulatory module	C ⁺	p-value^a	C ^–	Rank^b	C ^R	Rank^b
1	0.92	3.8E-23	0.0181	1	0.0155	1
2	0.92	8.8E-14	0.0776	8	0.0813	7
3	1	2.2E-20	0.0350	3	0.0308	4
4^c	1	6.5E-19	0.0362	5	0.0399	6
5	1	4.1E-19	0.0371	6	0.0385	5
6	1	3.4E-23	0.0360	4	0.0187	2
7	1	3.5E-14	0.1101	10	0.0922	9
8	1	1.9E-09	0.2018	11	0.2137	11
9	1	1.6E-13	0.1009	9	0.1035	10
10	0.77	5.9E-19	0.0200	2	0.0223	3
11	1	6.8E-15	0.0642	7	0.0813	8
Add1	1	5.5E-31	0.0091		0.0047
Add2	1	3.6E-47	0.0000		0.000267^d

^ap-values are calculated as probability of 13*C⁺ successful hits out of 13 trials, with probability of success in one trial C^R.
^bRank is according to the ratio C⁺/C^– and C⁺/C^R respectively.
^cFunctional module 4 does not rank high according to its p- value. One can easily find motifs combinations exhibiting superior statistics (for example, Add1 (Meme1 + Meme4 + TGIF) or Add2 (Meme1 + Meme2 + Meme5).
^dTo calculate this value, another random sequence set was generated by repeating 10 times the procedure of random splitting the Salmonella genome. In total the module can be found on 21 of such sequences.

ISSN: 1471-2105