Learning determinants of nucleosome positioning from sequence-based models of cell-free DNA

Placeholder Show Content

Abstract/Contents

Abstract
The human genome exists in cells as chromatin, a vastly complex 3D macromolecule with multiple hierarchical scales of structural and functional organization. Nucleosomes, small, radially symmetric protein octamers ~10-20nm in diameter, are the fundamental unit of chromatin organization from which higher order chromatin structures are derived. Nucleosomes act like passive beads on a string, controlled both by their own sterics and by higher-order processes such as TF binding and chromatin remodeling. This work centers on the relationship between nucleosomes and the DNA sequences they bind, measured using the mapping coordinates of (largely nucleosome bound) fragments of DNA isolate from blood plasma. In the first half, we develop a model called cfDNA-DeepNuc, which is a fully dense convolutional neural network trained to predict nucleosome occupancy from DNA sequence. We use the model as a basis to study the sequence determinants of nucleosomes positioning, identifying patterns that both coincide with and that extend beyond existing knowledge of nucleosome sequence determinants. In the second half, we present an approach to measuring chromatin state (i.e., enhancer accessibility) from cell-free DNA fragmentation patterns, and demonstrate proof of principle for providing estimates of tissue signal strength in a series of in silico titration and subsampling experiments. Our work is intended to highlight cell-free DNA fragmentation patterns as a functional genomic signal providing insight on genome structure and function, and to highlight the promise of using functional genomic annotation priors for the analysis of cell-free DNA in liquid biopsies.

Description

Type of resource text
Form electronic resource; remote; computer; online resource
Extent 1 online resource.
Place California
Place [Stanford, California]
Publisher [Stanford University]
Copyright date 2021; ©2021
Publication date 2021; 2021
Issuance monographic
Language English

Creators/Contributors

Author Probert, Christopher
Degree supervisor Curtis, Christina
Degree supervisor Kundaje, Anshul, 1980-
Thesis advisor Curtis, Christina
Thesis advisor Kundaje, Anshul, 1980-
Thesis advisor Greenleaf, William James
Thesis advisor Sidow, Arend
Degree committee member Greenleaf, William James
Degree committee member Sidow, Arend
Associated with Stanford University, Department of Genetics

Subjects

Genre Theses
Genre Text

Bibliographic information

Statement of responsibility Christopher Probert.
Note Submitted to the Department of Genetics.
Thesis Thesis Ph.D. Stanford University 2021.
Location https://purl.stanford.edu/fz292zp8053

Access conditions

Copyright
© 2021 by Christopher Probert
License
This work is licensed under a Creative Commons Attribution 3.0 Unported license (CC BY).

Also listed in

Loading usage metrics...