Abstract
Understanding the genetic basis of both rare and common diseases remains a major challenge. Despite advances in genomics, over half of individuals with rare diseases still lack a genetic diagnosis. This diagnostic gap suggests that conventional approaches may miss complex or unconventional forms of variation, such as tandem repeats (TRs)—sequences of repeated DNA motifs that vary in structure and length.
TRs are highly mutable and historically dismissed as "junk DNA," yet they are now implicated in over 60 rare diseases, including Fragile X syndrome and Huntington’s disease. Increasing evidence also suggests that TRs regulate gene expression and have biological function. However, due to the limitations of short-read sequencing, the variability and clinical relevance of most TRs remain poorly characterized. With the advent of long-read sequencing, it is now possible to study these regions at high resolution and scale. In this dissertation, I analyze TR variation across more than 1,000 long-read human genomes from the All of Us Research Program, aiming to map TR polymorphism and investigate its roles in disease.
To do this, I built a population-scale control database of TRs and developed novel computational tools for analyzing their genetic and functional complexity. Using TRGT, I generated high-resolution TR genotypes from PacBio HiFi data and benchmarked them against short-read ExpansionHunter calls. I profiled known pathogenic loci and compared them to non-pathogenic loci, developing new methods to decompose, visualize, and quantify TR sequence structure. These tools enabled discoveries in non-coding pentanucleotide TRs, an emerging class of pathogenic variation.
Together, these findings provide foundational resources and insights for the study of tandem repeats in rare disease, common traits, and human evolution.