Skip to main content

Table 1 Comparative analysis of TF binding specificity estimated based on three methods. Specificity threshold estimates by the uniform noise model [15], by ChIP -qPCR [15] and by the best-fit GDP function. The uniform random model-based estimations of the threshold values defined at FDR <5%. TF binding specificity threshold t was estimated based on the best-fit double truncated GDP function. The last two columns of the table shows that GDP-model could improve the specificity estimates providing by the Poisson model and ChIP-q-PCR assay.

From: Statistics of protein-DNA binding and the total number of binding sites for a transcription factor in the mammalian genome

TF

Unique mapped fragments, M*

Noise threshold by Poisson model*

# q-PCR experiments*

Noise 3-fold enrich threshold by qPCR*, q

Specificity threshold by GDP, t

at t

N2 at q*

at q

Specificity by qPCR, %

GDP-predicted specificity at q

Essrb

3609843

5

94

12

12

21646

21646

21600

94

99

Nanog

8424102

7

37

11

11

10343

10343

10213

97

98

oct4

4911144

6

47

8

11

1697

3761

2942

95

78

Sox2

4821446

6

48

8

13

2082

4526

5196

97

85

E2F1

8787961

6

48

9

16

13741

20699

21122

100

98

Tcfcp2I1

8449181

6

47

9

17

16293

26910

27912

97

96

ZFX

3844429

5

52

9

11

7161

10338

10966

100

94

Klf4

3807970

5

47

8

10

7433

10875

12122

97

88

c-Myc

6637404

7

48

9

12

1980

3422

2632

100

77

n-Myc

4823212

6

46

8

13

3214

7182

8780

97

78

STAT3

5351116

6

48

8

11

1229

2546

1983

97

78