Figure 3

Comparison of transcription factor binding site distributions in random sequence datasets and real promoter sequence dataset. Each curve represents the occurrence frequency of predicted binding sites in different data sets of comparable size. "Real Random Seq" is a data set of completely random sequences in which the emission probabilities of A, C, G and T were equal and uniform across the entire 2 kb. "Adjusted Random Seq" is a data set of random sequences generated with the adjusted emission probabilities of A, C, G and T according to that in the corresponding position at the real promoter sequence. "Real Seq" is the real promoter sequence dataset. An expectation value of 0.1 was used for detecting transcription factor binding sites in these datasets.