{"id":{"repo_id":"msu","oai_identifier":"oai:d.lib.msu.edu:etd_51688"},"canonical_url":"https://search.ndltd.org/etd/msu/oai:d.lib.msu.edu:etd_51688","repository":{"repo_id":"msu","name":"Michigan State University","base_url":"https://d.lib.msu.edu/oai"},"display":{"title":"APPLICATION OF DEEP GENERATIVE MODELING TO SINGLE CELL RNA SEQUENCING DATA IN TOXICOLOGY","abstract":"Thesis (Ph.D.)--Michigan State University. Pharmacology and Toxicology-Environmental Toxicology - Doctor of Philosophy, 2024","abstract_html":"Thesis (Ph.D.)--Michigan State University. Pharmacology and Toxicology-Environmental Toxicology - Doctor of Philosophy, 2024","abstract_has_math":false,"creators":["Kana, Omar"],"institution":null,"degree_name":null,"degree_level":null,"degree_discipline":null,"degree_department":null,"school":null,"contributors":["Bhattacharya, Sudin","Kaminski, Norbert","Rockwell, Cheryl","Nault, Rance"],"advisors":[],"committee_chairs":[],"committee_members":[],"year":2024,"date_issued":"2024","date_published":"2024","updated_at":"2026-11-18T18:58:38Z","subjects":["Bioinformatics","Toxicology"],"languages":["eng"],"rights":["Attribution-ShareAlike 4.0 International"],"rights_urls":[],"identifier_entries":[{"key":"dc:identifier","label":"Identifier","values":["etd:51688","local:Kana_grad.msu_0128D_20516"],"render_values":[{"text":"etd:51688","href":null,"code":true},{"text":"local:Kana_grad.msu_0128D_20516","href":null,"code":true}]}],"person_identifier_entries":[]},"links":{"outbound_url":"https://doi.org/doi:10.25335/ta02-s163","outbound_label":"DOI","outbound_source":"dc:identifier","record_url":"https://doi.org/doi:10.25335/ta02-s163","record_label":"Repository record","record_kind":"doi","record_source":"dc:identifier","download_url":null,"download_label":null,"download_source":null,"access_status":null,"embargo_until":null,"access_note":null},"source_record":{"url":"https://d.lib.msu.edu/oai?verb=GetRecord&metadataPrefix=oai_dc&identifier=oai%3Ad.lib.msu.edu%3Aetd_51688","prefix":"oai_dc"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor","label":"Contributor","values":["Bhattacharya, Sudin","Kaminski, Norbert","Rockwell, Cheryl","Nault, Rance"]},{"key":"dc:creator","label":"Author","values":["Kana, Omar"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date","label":"Dc Date","values":["2024"]},{"key":"dc:relation","label":"Dc Relation","values":["Electronic Theses & Dissertations"]},{"key":"dc:type","label":"Dc Type","values":["text","Electronic dissertations"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["Bioinformatics","Toxicology"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language","label":"Dc Language","values":["eng"]},{"key":"dc:rights","label":"Dc Rights","values":["Attribution-ShareAlike 4.0 International"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier","label":"Identifier","values":["etd:51688","local:Kana_grad.msu_0128D_20516","https://doi.org/doi:10.25335/ta02-s163"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description","label":"Description","values":["Thesis (Ph.D.)--Michigan State University. Pharmacology and Toxicology-Environmental Toxicology - Doctor of Philosophy, 2024","Single cell RNA-sequencing provides an opportunity for pharmacologists and toxicologists study the heterogeneity of cellular responses to chemical perturbations. However, analyzing these datasets remains incredibly challenging as they are high volume and high dimensional. Efforts to mitigate these challenges have centered on dimensionality reduction. Recently, it has been shown that variational autoencoders (VAEs), deep generative models that can perform dimensionality reduction, can be successfully deployed on single cell RNA-seq data to perform normalization and prediction. In this thesis I will explore both use cases.Even with the unprecedented detail single cell RNA-seq can provide for describing cell type specific chemical responses, exploring all relevant combinations of cell type-chemical perturbations remains difficult. Additionally, the dose of a chemical changes the overall character of its response. Variational autoencoders (VAEs) have been shown to predict chemical perturbations for single doses. However, VAEs have yet to be used to predict the entire dose-response. Here I introduce single cell Variational Inference of Dose-Response (scVIDR) which not only predicts the trajectory of the dose-response, but also achieves better large dose predictions than previous VAE algorithms. First, scVIDR is shown to predict dose-dependent gene expression across cell types in mouse liver, human blood cells, and cancer cell lines. Next, regression on scVIDR’s latent space is used to biologically interpret model predictions. Finally, scVIDR is used to order individual cells based on their chemical sensitivity by assigning a pseudo-dose value. I conclude scVIDR can effectively be used to predict chemical perturbations in a wide range of administration scenarios. Analysis of the compartmentalization of liver metabolism can be described with two axes: spatial and temporal. The spatial axis is conferred by the hepatic lobule, which is made up of concentric layers or zones of cells that have distinct metabolic programs. The master regulatory pathway that sustains this metabolic program is the Wnt/β-catenin pathway, which when activated by Wnt ligands, induces the transcription of potentiators of metabolic and nutritional gradients. The temporal axis of liver is conferred by the circadian rhythm which originates from super chiasmatic nucleus. Within each cell in the liver, this circadian rhythm is maintained by a core set of genes which confer feedback loops which sustain the necessary oscillations for metabolic efficiency. Previous studies have shown how the axes of the liver lobule can interact using single cell RNA-seq data. However, how these axes interact with toxicological perturbation is still unknown. In this thesis, a variational autoencoder is used to batch correct single cell RNA-seq data for zonation inference. Existing models originally used to analyze the zonal-rhythmic axes of the liver lobule are extended to account for the effects of chemical perturbation. This methodology is applied to mouse liver snRNA-seq data from mice subjected to acute treatment of 2,3,7,8 tetrachlorodibenzo-p-dioxin.","Description based on online resource. Title from PDF t.p. (Michigan State University Fedora Repository, viewed ).","Includes bibliographical references."]},{"key":"dc:format","label":"Dc Format","values":["application/pdf","1 online resource (130 p.) : ill."]},{"key":"dc:title","label":"Title","values":["APPLICATION OF DEEP GENERATIVE MODELING TO SINGLE CELL RNA SEQUENCING DATA IN TOXICOLOGY"]}]}],"canonical_facts":{"dc:contributor":["Bhattacharya, Sudin","Kaminski, Norbert","Rockwell, Cheryl","Nault, Rance"],"dc:creator":["Kana, Omar"],"dc:date":["2024"],"dc:description":["Thesis (Ph.D.)--Michigan State University. Pharmacology and Toxicology-Environmental Toxicology - Doctor of Philosophy, 2024","Single cell RNA-sequencing provides an opportunity for pharmacologists and toxicologists study the heterogeneity of cellular responses to chemical perturbations. However, analyzing these datasets remains incredibly challenging as they are high volume and high dimensional. Efforts to mitigate these challenges have centered on dimensionality reduction. Recently, it has been shown that variational autoencoders (VAEs), deep generative models that can perform dimensionality reduction, can be successfully deployed on single cell RNA-seq data to perform normalization and prediction. In this thesis I will explore both use cases.Even with the unprecedented detail single cell RNA-seq can provide for describing cell type specific chemical responses, exploring all relevant combinations of cell type-chemical perturbations remains difficult. Additionally, the dose of a chemical changes the overall character of its response. Variational autoencoders (VAEs) have been shown to predict chemical perturbations for single doses. However, VAEs have yet to be used to predict the entire dose-response. Here I introduce single cell Variational Inference of Dose-Response (scVIDR) which not only predicts the trajectory of the dose-response, but also achieves better large dose predictions than previous VAE algorithms. First, scVIDR is shown to predict dose-dependent gene expression across cell types in mouse liver, human blood cells, and cancer cell lines. Next, regression on scVIDR’s latent space is used to biologically interpret model predictions. Finally, scVIDR is used to order individual cells based on their chemical sensitivity by assigning a pseudo-dose value. I conclude scVIDR can effectively be used to predict chemical perturbations in a wide range of administration scenarios. Analysis of the compartmentalization of liver metabolism can be described with two axes: spatial and temporal. The spatial axis is conferred by the hepatic lobule, which is made up of concentric layers or zones of cells that have distinct metabolic programs. The master regulatory pathway that sustains this metabolic program is the Wnt/β-catenin pathway, which when activated by Wnt ligands, induces the transcription of potentiators of metabolic and nutritional gradients. The temporal axis of liver is conferred by the circadian rhythm which originates from super chiasmatic nucleus. Within each cell in the liver, this circadian rhythm is maintained by a core set of genes which confer feedback loops which sustain the necessary oscillations for metabolic efficiency. Previous studies have shown how the axes of the liver lobule can interact using single cell RNA-seq data. However, how these axes interact with toxicological perturbation is still unknown. In this thesis, a variational autoencoder is used to batch correct single cell RNA-seq data for zonation inference. Existing models originally used to analyze the zonal-rhythmic axes of the liver lobule are extended to account for the effects of chemical perturbation. This methodology is applied to mouse liver snRNA-seq data from mice subjected to acute treatment of 2,3,7,8 tetrachlorodibenzo-p-dioxin.","Description based on online resource. Title from PDF t.p. (Michigan State University Fedora Repository, viewed ).","Includes bibliographical references."],"dc:format":["application/pdf","1 online resource (130 p.) : ill."],"dc:identifier":["etd:51688","local:Kana_grad.msu_0128D_20516","https://doi.org/doi:10.25335/ta02-s163"],"dc:language":["eng"],"dc:relation":["Electronic Theses & Dissertations"],"dc:rights":["Attribution-ShareAlike 4.0 International"],"dc:subject":["Bioinformatics","Toxicology"],"dc:title":["APPLICATION OF DEEP GENERATIVE MODELING TO SINGLE CELL RNA SEQUENCING DATA IN TOXICOLOGY"],"dc:type":["text","Electronic dissertations"]},"updated_at":"2026-11-18T18:58:38Z"}