Affiliations: Computational Biology Branch, National Center for
Biotechnology Information, National Library of Medicine, National Institutes of
Health, Bethesda, MD, USA
Note: [] Corresponding author: Computational Biology Branch, National
Center for Biotechnology Information, National Library of Medicine, National
Institutes of Health, Building 38A, Room 8N805, Bethesda, MD 20894, USA, Fax:
+1 301-435-7794, E-mail: [email protected]
Note: [] Present address: Institute of Evolution, University of Haifa,
Haifa, 31905, Israel
Abstract: Sequence-dependent DNA curvature is known to play an important role
in initiation of transcription of many genes. We compared the
distribution of predicted intrinsic curvature of Escherichia coli and human
promoter sequences with the distribution of curvature of randomly selected
coding and non-coding fragments from these organisms. Different
methods of curvature calculation were found to yield mostly similar overall
tendencies of DNA curvature in all groups of sequences. According to
all methods of calculation, E. coli promoters were found to be more curved than
coding sequences from the same genome and random sequences with the same
nucleotide composition. By contrast, the average curvature of human
promoter sequences was only marginally greater than the average curvature of
human coding sequences. Non-coding intron sequences were found to be
the most curved of the human sequences examined. Based on these
observations, we hypothesize about the role of DNA curvature in promoter
sequences.
Keywords: DNA curvature, bendability, sequence analysis
