Author ORCID Identifier
0000-0001-9754-0967
Date of Graduation
5-2025
Document Type
Dissertation (PhD)
Program Affiliation
Quantitative Sciences
Degree Name
Doctor of Philosophy (PhD)
Advisor/Committee Chair
Ken Chen
Committee Member
Ziyi Li
Committee Member
Jeffrey T. Chang
Committee Member
Rui Chen
Committee Member
Nicholas Navin
Committee Member
Jichao Chen
Abstract
During the development of multicellular organisms, individual cells make distinct decisions about their cell types and states. Understanding the molecular mechanisms underlying cellular state transitions at different developmental stages provides deep insights into physiology, morphology and the etiology of diseases. Single-cell RNA-sequencing (scRNA-seq), which is widely used to study complex cell states and dynamic gene expression patterns, enables us to investigate molecular mechanisms of cellular state transitions. Currently, however, computational tools available for identifying cellular states and state transitions remain limited.
Although trajectory-based methods such as Monocle and Slingshot assume that state transitions generate continuous expression profiles, they cannot distinguish transition cells from stable ones. Cellular state transitions, which initiate key steps of developmental processes such as differentiation, dedifferentiation, and transdifferentiation, ultimately determine distinct cell fates. While methods like CellRank and MuTrans identify transition cells using macrostates and attractor basins, they rely on cell-cell similarity rather than intrinsic gene regulatory mechanisms. A systematic approach for distinguishing and characterizing transition cells from stable cells is still lacking.
In this dissertation, I model scRNA-seq data from developmental processes as a function of gene regulatory relations using stochastic differential equations (SDEs). Based on dynamical systems theory, I developed a statistical approach, CellTran, which leverages pairwise gene expression correlation coefficients to infer cell state transitions. I validated this method using simulation datasets generated by SERGIO and real-world mouse tissue regeneration scRNA-seq data. I also applied the analytical framework to scRNA-seq datasets from a PDAC (Pancreatic ductal adenocarcinoma) mouse model and humoral responses against SARS-CoV-2 vaccines. By identifying transition cells in pancreatic preinvasive lesions, we observe tumor heterogeneity and can predict distinct cell fate even at early stages of tumorigenesis. Additionally, characterizing transition cells after SARS-CoV-2 vaccination allows us to observe the protective effects of vaccines against severe cases. Identifying and characterizing transition cells can potentially provide valuable biomarkers for disease diagnosis, prognosis and therapeutic development.
Recommended Citation
Wang, Yuanxin, "Identifying and Characterizing Transition Cells in Developmental Processes from scRNA-Seq Data" (2025). Dissertations & Theses (Open Access). 1435.
https://digitalcommons.library.tmc.edu/utgsbs_dissertations/1435
Keywords
Cellular State Transitions, Dynamic Systems, Cell Development, Carcinogenesis, Single-Cell RNA Sequencing, Differential Equations, Gene Expression Correlation, Statistical Analysis