TF binding sites (cont.)
only 19% of noncoding sequence has >50%
probability of being aligned, but this 19% contains
74/75 of the muscle-specific sites and 75% of the
Sp1 binding sites
if the Gibbs sampler looks for motifs in the 19% of
conserved sequence, the three principal motifs are
found easily
conclusion: comparative genomics may be a fruitful
approach to detection of regulatory sites in DNA