Cross-species transfer learning of gene expression programs in tumor microenvironment
No Thumbnail Available
Authors
Meeting name
Sponsors
Date
Journal Title
Format
Thesis
Subject
Abstract
Dogs have emerged as superior models for studying cancer immunotherapy compared to traditional mouse models due to several key advantages, including their spontaneous tumor development, intact immune systems, and genetic, physiological, and environmental similarities to humans. These characteristics make dogs particularly well-suited for exploring tumor-immune interactions and assessing immunotherapy strategies in a translationally relevant context. To fully leverage this potential, it is crucial to characterize and compare the molecular and immunological landscapes of canine and human tumor microenvironments (TMEs). With the long-term goal of uncovering both unique and shared molecular processes within the canine and human TMEs to inform immunotherapy strategies, we propose a novel framework for identifying gene expression programs (GEPs) in better annotated human scRNA-seq data and projecting them onto dog cancer datasets that enables robust cross-species learning. This dissertation addresses three primary objectives: (1) Identifying methodological challenges in cross-species analysis between canine and human cancer scRNA-seq data, (2) Exploring machine learning methods suitable for such analysis, and (3) Developing a framework that incorporates best practices for learning GEPs from human data and projecting them onto canine data. Our results reveal both conserved and species-specific molecular programs, offering new insights into tumor progression and immune dynamics across species. This dissertation provides a foundation for advancing comparative oncology and translational immunotherapy research by enabling more biologically grounded and scalable cross-species analyses.
Table of Contents
DOI
PubMed ID
Degree
PhD