UC San Diego

Chromatin is the complex of genomic DNA and proteins that fills the nucleus of eukaryotic cells. The packaging of chromatin within the nucleus involves hierarchical folding at several levels. The first two levels produce nucleosomes and the 30-nm chromatin fiber. Additional levels give rise to the higher order organization of chromatin. Although the details of this organization remain unclear, their elucidation has recently been facilitated by chromosome conformation capture (3C) and related experimental methods, such as Hi-C, which detect physical interactions between different loci of genomic DNA. Knowledge of these interactions may be used to infer the conformation of the chromatin fiber. Motivated by this possibility, the present dissertation proposes an overall plan to develop and validate computational methods for the analysis of DNA interactions from Hi-C experiments. Within this plan, solutions for three specific problems are described. The first problem involves deducing an ensemble of chromatin conformations consistent with a given set of reference contact probabilities (CPs). The proposed solution consists of matching the given reference CPs with CPs estimated from simulations of a bead-chain polymer model representing the chromatin fiber. The second problem concerns accelerating the estimation of CPs from the simulated conformation ensembles. The proposed solution relies on fitting the extended generalized lambda distribution (EGLD) to distributions of simulated internal distances. The third problem involves quantifying the specificity of DNA cleavages in Hi-C experiments. The proposed solution consists of expressing an empirical local site distribution, obtained from Hi-C data, as a linear combination of conditional local site distributions, obtained from a reference genomic sequence, and then inferring cleavage fractions as weights of the linear combination. The above computational methods are promising steps toward the analysis of DNA interactions to investigate the higher order spatial organization of chromatin. This dissertation also includes a review on knots in biophysics, and describes a computational approach to identify efficient target sites for trans- splicing ribozymes