Using comparative genomics to locate TF sites
	eukaryotic TF sites are "weak" and are lost in the noise--hard to detect with motif finders hypothesis--conserved noncoding regions would be enriched in TF sites 28 proteins with rat homologs, all expressed in muscle, with 75 known muscle binding sites in the 5' regions plus 24 Sp1 sites Gibbs unable to find any site in upstream regions human-rat alignment--only 19% showed good alignment (Bayes aligner)