engleski

Computational analysis of the location of highly curved DNA regions in human genome

Protein coding is generally considered to be the most important function of a DNA molecule inside a living organism. Still, only about 1% of DNA includes coding regions . The rest consists of non-coding regions involved in processes like DNA packaging and gene regulation where DNA flexibility is an important factor. In our work we refer to the interrelation between protein coding regions and sequence dependent DNA curvature of human genome. Our hypothesis is that non-coding, regulatory regions use additional structure-encoded information as a key functional component and thus are more likely to contain highly curved regions, as opposed to exonic DNA serving primary as sequence-encoded information carrier. We created a whole-genome curvature profile of the human genome (data taken from Genome Browser, golden path release hg18 ) by allocating predicted curvature score to each nucleotide of all chromosomal DNA sequences. We used a data mining method called “ association rules” , to test the co-location frequency of highly curved areas and exons. It was shown that the occurrence of highly curved regions in exons was around two times less frequent than what would normally be expected. Association rules method was applied on truncated chromosomes obtained by removing nucleotides that were simultaneously non-coding and low curved, further showing an extremely low co-occurrence between highly curved and exonic regions. Correlation of curvature magnitude with human genome coding region positions in truncated chromosomes was shown to be as low as -0, 98. Our results demonstrate a specific, opposing relationship between coding regions and curved DNA, which is another step towards understanding how structural features in DNA relate to its functional aspects.

DNA; curvature; introns; exons

nije evidentirano