Graduation Date

Spring 5-9-2026

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Programs

Molecular Genetics & Cell Biology

First Advisor

M. Jordan Rowley

Abstract

The three-dimensional organization of the genome is a fundamental basis of gene regulation. Chromatin is partitioned into transcriptionally active and inactive compartments, and changes in this organization have been linked to disease, developmental transitions, and cell-type-specific gene expression. Hi-C has enabled genome-wide mapping of these compartments, but the standard computational approach, eigenvector decomposition of the Hi-C contact matrix, imposes resolution constraints that systematically obscure compartment structure at biologically relevant scales. Achieving sub-kilobase resolution with this method requires sequencing depths that often exceed 10 billion contacts, making high-resolution compartment analysis prohibitively expensive and inaccessible for most studies.

This dissertation introduces CRUSH (Compartment Refinement for the Ultraprecise Stratification of Hi-C), a hierarchical multi-resolution algorithm that measures compartment identity by directly quantifying each genomic locus's interaction affinity for active versus inactive chromatin environments, enabling reliable compartment detection at 1 kilobase resolution with fifty-fold less sequencing depth than eigenvector-based methods require.

Benchmarked across lymphoblastoid cell lines, ENCODE cell lines, meiotic germline cells, and triple-negative breast cancer lines, CRUSH consistently outperforms conventional approaches and recovers regulatory-element-level compartment information validated against nascent transcription and histone data. Systematic analysis of seven ENCODE cell lines reveals a monotonic, dose-dependent relationship between active histone mark co-occupancy and compartment score and demonstrates that H3K27me3 adopts context-dependent rather than constitutively repressive (compartmental) behavior in leukemia cells, suggesting that chromatin compartmentalization is more plastic and cell-type-specific than previously understood.

Applied to sparse single-cell Hi-C data, CRUSH enables per-cell compartment profiling at 25 kilobase resolution, achieving superior cell-type separation relative to PCA-based methods and recovering tissue-specific gene regulatory programs from individual cells without requiring aggregation of individual Hi-C maps into one bulk Hi-C map. Extended to 52,000-year-old woolly mammoth PaleoHi-C data, CRUSH identifies genomic regions with differential compartmentalization between mammoth and Asian elephant skin, including hair follicle developmental regulators consistent with the woolly mammoth's distinctive skin.

Collectively, this work establishes that chromatin compartments are fine-scale, locally determined features whose apparent large size in conventional analyses is an analytical artifact rather than biological reality. By overcoming the sequencing depth and computational barriers that have long constrained the field, CRUSH makes high-resolution compartment analysis routine and broadly accessible through an open-source implementation, expanding the scope of three-dimensional genome studies across development, disease and evolution.

Rights

The author holds the copyright to this work and any reuse or permissions must be obtained from the author directly.

Available for download on Tuesday, April 27, 2027

Share

COinS