This dissertation, comprising three projects, presents Bayesian statistical methods for analyzing heterogeneous multivariate data, with application to marker expression data obtained from cytometry at time-of-flight (CyTOF). In the first project, a Bayesian feature allocation model (FAM) is presented for identifying cell subpopulations based on multiple samples of cell surface or intracellular marker expression level data obtained by CyTOF. Cell subpopulations are characterized by differences in expression patterns of markers, and individual cells are clustered into the subpopulations based on the patterns of their observed expression levels. A finite Indian buffet process is used to model subpopulations as latent features, and a model-based method based on these latent feature subpopulations is used to construct cell clusters within each sample. Non-ignorable missing data due to technical artifacts in mass cytometry instruments are accounted for by defining a static missingship mechanism. The second project builds upon the first by introducing a repulsive FAM (rep-FAM) which restructures the probability distribution of a traditional FAM to identify features more likely to be distinct from each other. The problem that a conventional FAM has a positive probability of repeating a feature is eliminated by the rep-FAM, which also increases the probability of larger differences between features. The rep-FAM thus yields clusters that are more biologically interpretable than those identified by a conventional FAM. The third project presents methods for differential distributions between two experimental conditions, in the context of CyTOF data. A zero-inflated mixture of log-skew-t distributions is used to model the multi-modal, heavy tailed, and often highly skewed distributions that arise from these marker expression levels. A distance metric is proposed to quantify the degree of difference between distributions under various experimental conditions. In each chapter, we explore the performance and limitations of our proposed methodologies through simulation studies and real data analyses.

Modern marriage has been characterized as a “greedy institution” where couples disengage from their social networks and community to devote their undivided attention to each other. I compare the co-presence patterns, that is, who people surround themselves with and how much time is spent with them, of cohabiters and married individuals to determine whether marriage is a uniquely “greedy institution” or whether co-residential romantic relationships lead to similar social worlds. I also compare divorcees and never-married individuals to determine whether experiencing marital dissolution influences the time that is allocated to others. Pooling data from the American Time Use Survey (ATUS) from 2003 to 2013, this study finds little differences in how cohabiters and married persons manage their relationships outside of their own union. Although cohabitation may be no less “greedy” of social ties than marriage, cohabiting couples spend more time alone together than marital couples. Divorcees and persons who have yet to marry also have similar social worlds. Household composition and demographic characteristics play a large role in the differences in time allocation. The findings suggest that marriage is no “greedier” than informal co-residential unions, and social networks appear to recover post-marriage.

Cell surface proteolysis plays a crucial role in regulating various biological and cellular processes, including but not limited to cell signaling, protein maturation, and extracellular matrix remodeling. Dysregulation of these proteolytic events is often a hallmark of numerous diseases. Understanding these changes in proteolysis may present opportunities for therapeutic intervention as neo-epitopes are created that may be preferentially displayed in diseased states. Under hypoxic conditions, Pancreatic Dunctal Adenocarcimoma (PDAC) experiences global changes to its cell surface proteolytic state. Additionally, it is important to understand the mechanism of action for protein-protein interactions in disease, particularly in the case of viral infection, as these interactions may be therapeutic targets for disruption. The following work describes the development of novel protein tools to understand how cell surface proteolysis changes when various oncogenes are overexpressed in breast cancer and how SARS-CoV-2 Spike protein interacts with the human ACE2 receptor. In Chapter 1, I describe the work done to develop a method utilizing subtiligase to study changes to the proteolytic landscape of MCF10A cells under the influence of KRAS G12V and Her2 overexpression. Chapter 2 describes the work done in studying the effect of hypoxia on cell surface proteolysis in Pancreatic Ductal Adenocarcinoma (PDAC). In Chapter 3, I describe the develop of various protein-Fc fusions to study the protein-protein interactions of Spike protein and ACE2.

Atomically thin transition metal dichalcogenides (TMDs, e.g., MoS2, MoSe2, WS2, WSe2) exhibit intriguing tunable properties. For instance, their electronic band structure and optical spectra can be significantly modulated under the application of an electric field. Such electrical tunability allows for interesting scientific research and versatile applications of the materials. In this dissertation, we investigate 2H-stacked bilayer WSe2 dual-gated devices encapsulated by boron nitride. Bilayer WSe2 hosts two competing low-lying excitons, namely the QK and QΓ intervalley excitons. Although its electronic band structure suggests the QK exciton as the lowest-lying exciton, we observe the QΓ exciton at ~18 meV below the QK exciton. Our observation reveals that the QΓ exciton has larger binding energy than the QK exciton. The QK and QΓ excitons possess different interlayer electric dipole moments, which give rise to different Stark shifts under the application of an out-of-plane electric field. By controlling the electric field, we can switch the energy ordering of the QΓ and QK excitons, and control which exciton dominates the luminescence of bilayer WSe2. Remarkably, both QΓ and QK excitons exhibit unusually strong two-phonon replicas, which are comparable to or even stronger than the one-phonon replicas. By detailed theoretical simulation, we infer the existence of numerous two-phonon scattering paths involving (nearly) resonant exciton-phonon scattering in bilayer WSe2. Such electric-field-switchable intervalley excitons with strong two-phonon replicas make bilayer WSe2 a distinctive valleytronic material.We have also observed coherent states between multiple excitons with different center-of-mass momenta in bilayer WSe2. By studying the reflection spectra, we resolve two weak features arising from KQ and KQ’ intervalley excitons. Under the application of the out-of-plane electric field, both excitons exhibit strong Stark splitting, coupled with each other and with the nearby KK intravalley exciton. We can quantitatively simulate the results with a model that considers defect-mediated coherent coupling between the KQ, KQ’ and KK excitons. The good experiment-theory agreement strongly supports the emergence of coherent states between momentum-mismatched excitons and provides important insight into the role of defects in coherent excitonic coupling in semiconductors.

Chimeric antigen receptor T cell (CAR T) therapy has shown extraordinary success in the treatment of hematological malignancies. However, its application to solid tumors has been largely limited by the immunosuppressive tumor microenvironment and toxicities. Because most CAR targets are tumor associated antigens, they are also expressed on normal tissues that are killed by CAR T cells in a process known as on-target, off-tumor toxicity. To mitigate this toxicity and increase specificity, Boolean logic gates have been applied to CAR T cells to regulate activity based on multiple antigenic signals. The AND-NOT-gate utilizes an inhibitory chimeric antigen receptor (iCAR) to downregulate T cell activation after normal tissue antigen recognition. Our study found that an iCAR with a single PD-1 domain shows a kinetic delay in inhibition. This delay can be ameliorated by increasing the avidity but not the affinity of the iCAR. Alternatively, the PD-1 domain can be replaced with alternative inhibitory domains from BTLA, LAIR-1, and SIGLEC-9. To further enhance inhibition, we developed a dual-signaling inhibitory CAR (DiCAR) that combines the PD-1 domain with one of these alternative domains. These DiCARs are more efficient at inhibiting cytotoxicity than an iCAR with a single PD-1 domain. Furthermore, preliminary studies have shown that these DiCARs can inhibit cytotoxicity in vivo.