Predicting RNA Structure and Discovering Functional RNA Sequences in Genomes Using Folding Thermodynamics

Thesis (Ph.D.)--University of Rochester. School of Medicine and Dentistry. Dept. of Biochemistry and Biophysics, 2008.RNA is more than just an information carrier that constitutes the intermediate stage between DNA and protein as suggested in the original Central Dogma of Biology. With the discovery of novel non-protein-coding functions of RNA molecules whose secondary and tertiary structure are crucial for their function, it is becoming increasingly important to understand the details of RNA folding. In this work, RNA folding thermodynamics have been used to predict stacking in RNA multibranch loops and to find novel non-coding RNA candidates in genome-wide scans. Here, nearest-neighbor folding free energy parameters are used to predict coaxial stacking configurations in RNA multibranch loops. The method uses a partition function calculation to predict coaxial stacking and achieves significant accuracy (66.7% sensitivity and 51.9% positive predictive value) when compared to RNA molecules with available high-resolution crystal structures. Dynalign, an algorithm for determining the most stable RNA secondary structure common to two sequences, has been used to search yeast genomes (S. cerevisiae and C. albicans) for unannotated regions with predicted high stability of secondary structure. A support vector machine is used to find regions that are predicted to be more stable than random background. Initial results suggest that some of the candidate regions are part of these organisms’ transcriptomes and are likely to be novel ncRNAs