UC Santa Cruz

Transfer RNAs (tRNAs) are universal to living organisms, and are involved in a variety of noncanonical functions. Systematically identifying atypical tRNAs is of considerable biological interest, but extracting relevant insights from a large trove of sequencing data is difficult due to the complexity of tRNA variation. This dissertation provides high-throughput tools to annotate, analyze, and visualize tRNA sequence features.

Using hundreds of thousands of predicted tRNAs and a multi-round strategy, I developed new isotype-specific covariance models across the three domains of life. These models accurately call a tRNA’s isotype given its sequence, highlight cases of isotype uncertainty, and help predict whether a tRNA exhibits translational function.

I also created a framework for comparative analysis of tRNA sequence features. Using this framework, I determined ubiquitous, isotype-specific, and clade-specific sequence features for over 150 eukaryotic species spanning across several clades. These consensus sequence features and their exceptions present new insights into tRNA function and evolution.

Taking full advantage of this framework, however, requires expert domain knowledge. To address this, I designed and built tRNAviz, an application that provides fast, interactive visualizations for sequence features of queried clades and isotypes, and allows users to compare uploaded sequences with groups of tRNAs. As a proof of concept, I conducted analyses of prokaryotic sequence features entirely in tRNAviz.

Together, these tools provide primary sequence context for tRNAs, illuminate cases of clade-specific tRNA evolution, and accelerate and democratize the discovery of atypical tRNAs.