UC Davis

RNA is a key biopolymer that mechanistically drives many cellular processes. As a single-stranded molecule with a flexible sugar-phosphate backbone, it can fold into intricate structural conformations. The functions, interactions, and regulations of RNA are often directly attributable to these structures; as such, understanding structure is crucial to deciphering the mechanisms of RNA function and dysfunction. High quality structure models can be obtained with nuclear magnetic resonance (NMR) and X-ray crystallography. However, these methods are low-throughput, encumbered by technological limitations, and lack applicability in vivo. In recent decades, structure profiling (SP) experiments have emerged as a practical and scalable approach to measure the structures of RNA transcripts in their in vivo contexts. These methods work by exposing transcripts to chemical reagents that induce covalent modifications in a structure-dependent manner. Modifications can be mapped by high-throughput sequencing, resulting in nucleotide-wise measurements of stereochemical characteristics. SP experiments have now scaled to the level of the entire transcriptome, enabling structure studies with unprecedented scope and depth. However, the data from these experiments have been largely underutilized due to a lack of computational tools capable of readily processing their massive scale when linking structure to function.

This dissertation focuses on the development of methods to interpret transcriptomic structure profiling data. I devise a novel statistical model of SP data and couple it to a data-driven structure recognition algorithm, yielding an accurate, automated, and scalable tool for identifying structures and structure-function relationships. Application of the method to diverse datasets demonstrates its utility in several domains. Specifically, it reveals novel insights on mRNA structure dynamics, characterizes structures within viral RNA genomes, profiles the RNA-protein interactome, and links specific structure motifs to post-transcriptional regulation. The method is adaptable for future types of profiling experiments and readily scales to the evolving scope of structure studies. Altogether, these results have helped further our understanding of in vivo RNA dynamics and provide the RNA community with a versatile tool to assess the transcriptomic structural landscape.