Scalable Integrative Frameworks for Multi-Omic Systems Biology
Date Issued
August 2026
Author(s)
Merlet, Jean
The University of Tennessee
Advisor(s)
Daniel Jacobson
Additional Advisor(s)
Helen Baghdoyan
Jon Beever
Michael Langston
Abstract
Network-based systems biology and scalable computational workflows can be used to integrate and interpret increasingly large and diverse biological datasets. Here, we investigate two biological case studies through this framework, spanning single-cell transcriptomics, spatial transcriptomics, viral genomics, and epidemiological data. First, in a study on atopic dermatitis, we integrate single-cell RNA sequencing and spatial transcriptomic datasets across murine and human systems using transcriptomic clustering, spatial neighborhood graphs, and multiplex random walk with restart diffusion. Using these approaches, we identify inflammatory fibroblast populations and immune-regulatory networks associated with eosinophilic inflammation. Second, we address the computational scalability challenges underlying modern, large-scale omics analyses through the development of HALRA, a sparse-preserving and distributed-memory implementation of the ALRA single-cell imputation algorithm designed to enable atlas-scale scRNA-seq analyses. Finally, we construct a large-scale, data-driven genomic similarity network through exhaustive comparisons of millions of whole-genome SARS-CoV-2 sequences, identifying over 22,000 biologically coherent viral lineages from the network topology using Markov clustering. We integrate these lineages with two years of weekly-resolution, county-level mortality and vaccination data by creating a spatiotemporal multiplex network. We investigate relationships between viral genomic structure and sociodemographic differences through random walk with restart diffusion and show lineage-specific spatial patterns associated with epidemiological variation.
Degree
Doctor of Philosophy
Major
Data Science and Engineering
File(s)![Thumbnail Image]()
Name
MERLET-PRIMARY-2026.pdf
Size
14.53 MB
Format
Adobe PDF
Checksum (MD5)
672303a8b15426443fd238753833f80b