Systems Interrogation of Host-Microbiome Immunomodulation and Metabolism

Abstract

In this dissertation, I interrogate how scaling computational systems for complex omics problems efficiently can lead to novel biological discoveries in the context of both how the microbiome modulates host metabolism and how the immune system responds to cancer. In Chapter 2, the method Aether is introduced. Aether is a tool that allows for intelligence bidding on cloud compute to reduce the cost of computational tasks in genomics by up to .7 orders of magnitude. In Chapter 3, Aether is utilized to handle de novo assembled metaomic data at massive scale to help yield the discovery a novel microbe present in the stomach of professional athletes. Chapter 4 shows how working with complex single cell sequencing data of immune cells at scale can yield novel insight into tumors non-invasively through the creation of machine learning algorithms that can predict whether a CD8+ T cell in blood is in a shared clonal lineage as T cells in tumor. Taken together, these projects demonstrate the power of coupling experimental design with computation at scale.