Date of Award


Degree Type


Degree Name

Doctor of Philosophy (PhD)



First Advisor

Michael Gribskov

Committee Chair

Michael Gribskov

Committee Member 1

Dan Goldwasser

Committee Member 2

Daisuke Kihara

Committee Member 3

Cynthia Stauffacher


Eukaryotic genomes are pervasively transcribed; almost every base can be found in an RNA transcript. This is a surprising observation since most of the genome does not encode proteins. This RNA must serve an important regulatory function – important because producing non-coding RNA is an energy intensive process, and in the absence of strong selection one would expect it to disappear.

RNA families with common functions have specifically conserved structural motifs, which are directly related to the functional roles of RNA in catalysis and regulation. Because the conserved structures depend on base-pairing, similar RNA structures may have little or no detectable sequence similarity, making the identification of conserved RNAs difficult. This is a particularly serious problem when studying regulatory structures in RNA. In many cases, such as that of cellular internal ribosome entry sites, although we can identify RNAs that have similar regulatory responses, it is difficult to tell whether the RNAs have common structural features using current methods. Available tools for identifying common structures based on RNA sequence suffer from one or more of the following problems: they do not consider pseudoknots, which are important in many catalytic and regulatory structures; they do not consider near minimum free energy structures, which is important as many RNAs exist as an ensemble of structures of nearly equal energy; they require many examples of known structures in order to train a computational model; they require impractical amounts of computational time, precluding their use on long sequences or genomic scale; or they use a similarity function that cannot identify RNAs as having similar structure, even when they are from one of the well characterized known classes. The approach presented here has the potential to address all of these issues, allowing novel RNA structures that are shared between RNAs with little or no sequence similarity to be discovered. This provides a powerful tool to investigate and explain the pervasive transcription observed in eukaryotic genomes.