Using hydroxyl radical cleavage data to study the DNA topography of non-coding functional sequences

It is difficult to decipher the functional signals that lie in the vast non-protein-coding landscape of eukaryotic genomes. The conventional method of mapping nucleotide sequence similarity has had limited success. While coding regions are easily identified with high specificity, many non-protein-coding functional regions are missed. Current methods are limited because they consider only the primary sequence of nucleotides. The three-dimensional structure of DNA is a critical feature recognized by the regulatory machinery within a cell. Because different nucleotide sequences can have the same local structure, it is possible for different sequences to be structurally similar. Given the critical nature of functional non-protein-coding DNA, the limited ability of traditional comparative sequence analysis techniques to detect these regions, and the prospect that structural features are conserved, it is likely that comparing DNA structural features will be useful. In this dissertation I use hydroxyl radical cleavage patterns as a measure of the local shape—or topography—of genomic DNA, at single nucleotide-resolution. First, I show that DNA structural motifs are detectable in non-protein-coding regions in the yeast genome that are bound by common transcription factors, even when primary sequence-based motifs are undetectable. Second, I explore the relationship between DNA topography and GC content, and show that regions of the human genome that have high cleavage patterns tend to overlap functional genomic elements. Third, I develop evolutionary constraint-detection algorithms that are informed by DNA topography. Using these algorithms, I find that about twice as much of the human genome is under evolutionary constraint compared to the amount detected by conventional sequence constraint algorithms. The additional structure-informed constrained regions correlate with functional non-proteincoding elements. Some of these regions act as transcriptional enhancers in cell culture experiments. Fourth, I show that larger changes in DNA topography (i) correlate with decreased binding affinity for transcription factors; and (ii) are enriched in phenotype-associated single nucleotide polymorphisms (SNPs). These results support the idea that the molecular shape and structure of DNA may be under evolutionary selection. I propose that considering local DNA structure in addition to nucleotide sequence will be important for gaining a comprehensive understanding of genome evolution and function.