Search

Background: Genome-wide association studies have identified numerous disease-associated variants, but a vast majority are located in non-coding regions, making it challenging to understand their functional impact. This complexity necessitates new techniques to identify causal variants in non-coding regions and elucidate their specific cellular contexts and mechanisms of action. Here we present work i) examining mutations that create nullomers in the human genome to explore its potential utility in identifying pathogenic mutations and ii) a single-cell multi-omic study identifying the transcriptome and regulome of the human and mouse hypothalamus to identify regulatory regions of obesity-associated variants.Methods: (i) We generated all possible mutations of the human genome that can lead to emergence of a nullomer, and examine where in the genome they emerge. (ii) We apply single-cell RNA and ATAC sequencing to adult hypothalamus samples.Results and Conclusions: (i) Our findings highlight CpG hypermutability and methylated cytosines as key elements leading to resurfacing of nullomers in individuals. We also showcase that nullomers can have applications in disease annotation and pathogenic variant identification. (ii) We identified regulatory elements of hypothalamus cell types and mapped obesity-associated variants to cell-type specific peaks. We validated these regions to be enhancers using CRISPR editing and CRISPRi.

Complex traits are intricately intertwined with an organism's genome, a relationship underscored by the dynamic landscape of its transcriptome. Selective gene expression regulates cell type specialization and fluctuation of cell states. The development of RNA sequencing has facilitated the capture of the whole transcriptome of a given sample. However, a bulk approach obscures cell type heterogeneity, impeding the precise dissection of cell-specific effects, including those modulated by genotype, developmental stage, and disease state. In contrast, single-cell and single-nucleus RNA-seq preserves cellular identity, enabling a comprehensive mapping of gene expression across various cell types and states.Here, I describe my work in single-cell transcriptomics to characterize cell types and cell states in mouse. First, I present our long-read single-cell RNA-seq method, benchmarked in the C2C12 mouse myogenic system, which revealed cell type-specific isoform switching in key genes during myogenesis. Next, I characterize 5 mouse tissues at single-nucleus resolution during postnatal development using the ENCODE4 mouse dataset, where I used topic modeling to reveal cell type- and state-specific cellular programs. Lastly, I investigate the impact of genetic variation on gene expression across 8 diverse tissues from 8 mouse genotypes, pinpointing genotype- driven variation in specific celltypes in both wild-derived and classical lab strains. Together, these projects lay the groundwork for cohesive cell type and cell state annotation and comparative analyses, contributing to future characterization of these tissues in other contexts such as human diseases and hybrid mouse genotypes.

Cancer has historically been considered a genetic disease originating from activating mutations of oncogenes or inactivating mutations of tumor suppressors. More recently, appreciation for the contributions of epigenetic mechanisms to carcinogenesis has grown as they help to explain the aberrant gene expression and cellular identity associated with many cancers. For example, changes in DNA methylation at CpG dinucleotides can alter gene expression to promote malignant properties. In pancreatic ductal adenocarcinoma (PDA), the genetic underpinnings of tumorigenesis are known and primarily originate from mutations in KRAS and TP53. Furthermore, transcriptomic studies have identified transcriptional programs that are commonly enriched in subsets of patients, resulting in two main subtypes of PDA with clinical implications: squamous-PDA and progenitor-PDA. An understanding of the genome-wide DNA methylation aberrations in PDA and their contribution to tumor progression and subtype identity is less clear. In this dissertation, the genome-wide DNA methylation changes across PDA stages and subtypes as well as the impacts on metastatic characteristics and subtype differentiation are explored. Chapter 1 is a review of the literature highlighting our current understanding of the DNA methylation landscape in PDA which has primarily been learned from microarray studies that assay a very limited subset of CpGs. In Chapter 2, we characterize the genome-wide DNA methylation landscape of PDA organoids using whole genome bisulfite sequencing (WGBS), identifying and characterizing stage-specific and subtype-specific DNA methylation differences that may be beneficial as prognostic and predictive biomarkers. In Chapter 3, we identify subtype-specific DNA methylation differences in commercially available PDA cell lines and dissect GATA6-mediated epigenetic contributions to progenitor-PDA subtype differentiation. GATA6 is a known master regulator of progenitor-PDA and we found that its binding motifs are enriched in hypermethylated regions of squamous-PDA. While restoration of GATA6 expression in squamous-PDA did upregulate the progenitor-PDA gene signature, it did not reverse the observed hypermethylation suggesting that GATA6 does not directly mediate DNA methylation of its binding regions. Instead, GATA6 may epigenetically regulate transcription of its target genes through its interaction with the SWI/SNF family of chromatin remodelers. Thus, the SWI/SNF complex may be a therapeutically actionable target for treating patients with the progenitor-PDA subtype. In Chapter 4, we perform a combined transcriptomic and DNA methylome analysis of mouse PDA organoid models to identify genes whose methylation-mediated silencing likely contributes to PDA metastasis. We found that hypermethylation of the Pdzrn3 promoter represses Pdzrn3 transcription and promotes metastatic characteristics including increased cell migration and a pro-migratory cell morphology. Chapter 5 summarizes the main findings of this work, applications of the research, and future directions.

Since its inception in the 1970s with Frederick Sanger's pioneering work on Sanger sequencing, DNA sequencing technology has undergone a remarkable evolution. Automated Sanger sequencing in the 1980s and 1990s significantly increased throughput and accuracy, while the 21st century saw the emergence of next-generation sequencing (NGS) technologies like Illumina, revolutionizing genomics research by enabling high-throughput sequencing (van Dijk et al. 2014). However, short-read sequencing techniques, including Illumina, have limitations in accurately resolving repetitive regions and complex structural variations within the genome, hindering comprehensive understanding. In response, third-generation sequencing technologies such as Pacific Biosciences' (PacBio) and Oxford Nanopore Technologies (ONT) have arisen, offering long-read capabilities but facing challenges like DNA polymerase limitations (PacBio) and high error rates (ONT)(Rhoads and Au 2015). Despite these challenges, both PacBio and ONT sequencing technologies continue to advance rapidly, with potential to deepen our understanding of the genome's complexities (Wang et al. 2021). Leveraging the advantages of ONT sequencing, and developing molecular biology methods and bioinformatics tools that overcome the limitations of ONT sequencing to address the unknowns of the adaptive immune system (AIS), is the focus of my graduate work. The diversity of antibody repertoires poses significant challenges for sequencing and annotation. By developing assays that overcome these challenges, my research contributes to a deeper understanding of antibody diversity across vertebrate species, with implications for antibody-based therapy and research applications.

Understanding the genomic architecture of speciation remains a challenge in evolutionary biology. Among broadcast-spawning marine invertebrates, reproductive isolation is thought to be established and maintained by the divergence of gamete recognition proteins located on the surfaces of sperm and egg cells. However, it remains unclear whether gametic isolation has been an effective barrier to gene flow during and/or following speciation. In this dissertation, I characterized the history of introgression among the North Pacific sea urchin species of the family Strongylocentrotidae to deepen our understanding of their diversification and evaluate the importance of gametic isolation in speciation. Using whole-genome sequencing data from each strongylocentrotid species and cutting-edge phylogenomic approaches, I documented widespread introgression in both extant taxa and ancestral lineages, demonstrating that gametic isolation did not effectively limit introgression. I implemented a phylogenetic hidden Markov model to locate the specific regions of the genome affected by introgression, finding evidence of strong selection against introgression across much of the genome. Although introgressed variation has predominantly persisted in slowly evolving, low-divergence genomic regions, numerous protein-coding genes showed both introgression and historical positive selection, suggesting an adaptive role for introgression. Finally, I showed that the two gamete recognition proteins responsible for species-selectivity in sea urchin fertilization, sperm protein bindin and its egg receptor, EBR1, have experienced historical adaptive introgression, a pattern inconsistent with expectations for barrier loci. My findings contribute to the body of literature evaluating the biological consequences of introgression and question the importance of gamete recognition proteins in the evolution of reproductive isolation among incipient strongylocentrotid sea urchin species.

U2AF1S34F, a somatic splicing factor mutation, is frequently recurrent in human neoplasias such as lung adenocarcinoma (ADC). Although U2AF1S34F has been shown to occur early in tumor lineages, the mutation, alone, is insufficient for producing tumors. However, lung ADC patients with U2AF1S34F frequently have co-occurring KRAS mutations and smoking histories. We hypothesized that U2AF1S34F interacts with oncogenic KRAS and environmental stress to promote tumor-forming potential. To elucidate interaction of U2AF1S34F with a co-occurring mutation, we generated human bronchial epithelial cells (HBEC3kts) with U2AF1S34F or with co-occurring U2AF1S34F and KRASG12V. From analyzing short-read transcriptome sequences, we found synergistic effects of co-occurring mutations on gene expression in cell cycle and inflammatory pathways associated with increased tumors in mouse xenografts, anchorage-independent growth, proliferation, and altered cytokine production.nterestingly, HBEC3kts harboring only U2AF1S34F display increased splicing in stress granule protein genesand increased viability in cigarette smoke concentrate. Our results suggest that U2AF1S34F may prime cells for transformation by allowing precancerous cells to survive longer when environmental stress is present, permitting U2AF1S34F cells to accumulate transforming mutations, such as KRASG12V.Next, I sought to further investigate the impact of U2AF1S34F and environmental stress response by profiling the mRNA modification landscape of U2AF1S34F and U2AF1WT HBEC3kts exposed to cigarette smoke concentrate (CSC) using Nanopore direct RNA sequencing (dRNA-seq). Preliminary results show that RNA modificatios in autophagy gene VAMP8 were associated with altered protein expression levels. We also show that CSC and the presence of U2AF1S34F both increase the number of RNA modifications in the transcriptome, with the highest number of modifications occurring in CSC treated U2AF1S34F HBEC3kts. We hypothesize that U2AF1S34F and CSC modify the RNA modification landscape in a synergistic way to increase oncogenic potential.

When divergent populations come into contact and interbreed, incompatible genotypes can result in decreased hybrid fitness and partial reproductive isolation. By studying patterns of gene flow between populations, we can identify the genetic, phenotypic, and ecological components of partial reproductive isolation. Using whole-genome sequence data generated from 1201 yellow-rumped warblers (Setophaga coronata coronata, S. c. auduboni, and their hybrids), we measure variation of a hybrid zone’s position in space and over time, identify genomic regions where gene flow is restricted, and compare the genetic basis of reproductive barriers to the genetic basis of plumage traits. We find that reproductive isolation is generated by the effects of many loci throughout the genome, with a large influence of the sex-chromosomes, that “barrier loci” cluster in some highly differentiated regions, and that gene flow and introgression are broadly asymmetric. Barrier loci show strong linkage disequilibrium, which augments the strength of the barrier to gene flow and may indicate the presence of genetic incompatibilities. Clines in plumage traits and their associated SNPs suggest strong selection against some hybrid phenotypes, although loci with the strongest associations with plumage color traits do not exhibit the strongest barrier to gene flow.

G protein coupled receptors (GPCRs) are the largest family of integral membrane receptor proteins, responsible for mediating numerous signaling pathways that are vital to a wide range of biological processes. As a result, over a third of FDA-approved drugs target GPCRs to treat a variety of conditions. Additionally, dysfunctional GPCR and G protein signaling contributes to several disease states, including cancer. Previous studies have shown that dysregulation of Gαs-coupled protein receptors have an important role in the immunosuppressive properties of the tumor microenvironment (TME). Despite this, the function of GPCRs in the context of cancer immunotherapy remains largely underexplored. Here, we used a bioinformatics analysis pipeline to identify and analyze the immune cells present in the TME of 42 different mouse tumor samples. With this information, we found several GPCRs with conserved upregulation in mouse and human tumor samples. Of these GPCRs, several are targets of FDA-approved drugs, providing several potential targets for future cancer immunotherapeutic drugs.

A major goal of the biomathematics discipline is to optimize mathematical models for biological processes. This optimization can take on various forms; finding the appropriate model that fits available data, allows for accurate inference, and is computationally feasible is no easy task and requires an understanding of both the biological processes at hand and the mathematics behind each potential model or algorithm. In this dissertation, I seek to understand how mathematical modeling choices affect our ability to understand human disease. I study infectious, cancerous, and polygenic disease from a variety of computational perspectives. First, I apply methods for differential sensitivity analysis in biological models for both cancerous and infectious disease spread. I compare prediction accuracy for existing first-order methods and propose a second-order method with enhanced flexibility both in terms of the model for which it is applied and the programming environment available. Second, I compare statistical approaches for uncovering genetics of complex disease in admixed populations, using likelihood ratio tests to understand how to incorporate local ancestry in genome wide association studies to achieve the highest power. Third, I utilize machine learning methods to reduce diagnostic delay for patients across the University of California Health system. I adapt a logistic regression model to find patients likely to have common variable immune deficiencies from one health system to five health systems. I also adapt this algorithm from the immunology realm to the cardiology realm to predict cardiac amyloidosis. Along the way, I use this context to study automated feature selection, longitudinal feature engineering, and observational bias in electronic health record data.

Autoimmune disease arises from dysregulation of the immune system, leading to its attack on the body's own tissues and organs. The clinical heterogeneity of these diseases arises from several sources, such as genetic predisposition, environmental triggers, and aberrant immune responses. One emerging area of interest is the role of transposable elements (TEs) in autoimmune disease pathogenesis because these self-nucleic acids can be mistakenly detected as foreign, which can trigger a chronic immune reaction.There is growing appreciation for the role of TEs in systemic lupus erythematosus (SLE) and studies have found differentially expressed TEs in SLE patients, which suggests a link between TE activity and disease mechanisms. Our work investigated TE expression in four immune cell types from SLE patients, revealing cell-specific and SLE subphenotype-specific differentially expressed TEs, with additional cell-type-specific TE associations in different ancestry groups. TE expression was also associated with host gene expression involved in antiviral and immune responses, supporting the hypothesis that TEs could activate the innate immune system and contribute to chronic inflammation and autoimmunity.This study underscores the importance of TEs in SLE heterogeneity and highlights the need for further exploration of TE expression in normal immune cells and functional studies to understand their role in SLE pathogenesis. Future work to study whether antiretroviral drugs could reduce expression of TEs and mitigate SLE symptoms is warranted, given the potential involvement of TEs in autoimmune disease pathogenesis

With advances in mass spectrometry, there is increasing demand for more effective and adaptive approaches to systematically extract biological insights from glycomic and lipidomic data. This thesis explores the application of Markov modeling in understanding two key cellular processes, N-glycosylation and lipid biosynthesis.In Chapter 2, we demonstrate that a Markov model of N-glycosylation network successfully captures intricate glycosyltransferase interactions by reproducing a set of glycoprofiles from glycoengineered CHO cells producing erythropoietin (EPO). We further validate the model parsimony and accuracy by predicting the glycoprofiles of other glycoengineered drugs from the trained models and their wildtype glycoprofiles. The results attest to the model’s ability to learn glycosyltranferase activities and substrate specificities. To increase the impact of this approach, we also develop GlycoMME to allow broader access to this modeling pipeline, presenting a promising direction for rational glycoengineering. Chapter 3 extends the modeling approach to lipid biosynthesis, introducing the Lipid Synthesis Investigative Markov Model (LipidSIM). As a low-parameter, biologically interpretable framework, LipidSIM proves powerful in leveraging the interdependency in lipidomic data and extracts and quantifies perturbations to lipid biosynthesis reactions, generating hypotheses directly testable by transcriptomic data. This method is showcased in 5 different scenarios with 3 different lipidomic datasets, and the results substantiate LipidSIM as a valuable tool for extracting insights from high-dimensional lipidomic data of different types. In Chapter 4, a Taguchi design is used to systematically characterize the impact of 15 media supplements on potential cellular phenotypes for CHO cells, especially N-glycosylation. The analytical pipeline allows disentangling the impact of individual supplements at different concentration levels with minimal numbers of experimental configurations. This approach answers the demand to find more flexible strategies for controlling N-glycosylation beyond genetic engineering. When applied in conjunction with the Taguchi design, our modeling framework has the potential to facilitate rapid customization of media for the growing market of glycoprotein drugs.This thesis encapsulates the innovative applications of probabilistic modeling in accounting for the biological dependency underlying omics data, offering insights into intricate cellular processes and motivating further exploration in actionable modeling of biological systems.

The study of gene expression, the abundance of RNA transcripts in cells, has played a critical role in expanding knowledge of the molecular underpinnings of cancer and the ways in which treatments affect it. Because genes operate in interlinking pathways with nuanced and context-dependent behavior, deriving the phenotypic implication of changes in abundances of individual genes can be challenging. For this reason, Gene Set Enrichment Analysis (GSEA) was developed. GSEA is a statistical method that quantifies the activation of pathways or processes as represented by a priori annotated gene sets and plays a critical role in constructing pathway-level understandings of cell states in diseases such as cancer. In the use of GSEA and the expansion of its companion database of gene sets, the Molecular Signatures Database (MSigDB), two challenges have emerged. First, some gene sets lack the properties of context-specificity and coordinate regulation, that is, not all their members be collectively more expressed in samples that have a specific phenotype to which the gene set should correspond. Second, the new technologies for measuring gene expression at single cell resolution yield data with different properties than the “bulk” gene expression data for which GSEA was initially developed. I will address the first challenge in Chapter 1, in which I propose a data-driven method for gene set refinement and show that this method yields new gene sets that are both more coordinately regulated and context-specific. I address the second challenge in Chapter 2, in which I present a characterization of the performance of the single sample version of GSEA (ssGSEA) in multiple datasets and propose adaptations which I will show produce enrichment scores that are more stable and certain in the single cell context. Finally, Chapter 3 is an application of gene set based pathway characterization to the problem of understanding the development of chemoresistance in ovarian cancer. Using a cell cycle-aware approach incorporating single sample GSEA, we propose a new framework for modeling the development of resistance to carboplatin. These analyses together lay the groundwork for improving the robustness of GSEA results, adapting it to emerging technologies for gene expression measurement, and applying it to address key challenges in the treatment of cancer.

Biomolecular signaling networks enable cells to decode signals from their environment into the appropriate responses needed for survival and maintaining biological function. Using principles derived from these systems, synthetic approaches such as optogenetics were developed to grant users with precise control over cellular activity. Nonetheless, it remains challenging to engineer synthetic signaling systems towards predefined dynamic behaviors. In this work, we present a quantitative framework that integrates multi-objective optimization, dynamic modeling, and biosensor imaging to streamline the design and implementation of synthetic signaling networks in cells. First, we formulated a general signaling model and applied an evolutionary algorithm to derive a network design capable of decoding two distinct frequency modulated stimuli into separate response channels. Based on this design, we implemented an optogenetic circuit to control protein interactions in cells according to the blue light input pattern. To aid this process, dimerization-dependent fluorescent proteins were used to characterize and select optogenetic components with the required dynamic characteristics. Lastly, we applied our frequency decoder system to regulate downstream biological outputs such as gene transcription, antigen presentation, and antigen triggered CAR-T killing in mice. This framework provides an integrative approach for developing synthetic signaling systems that reduces reliance on trial-and-error.

Insulin resistance is implicated in multiple prevalent metabolic disorders, such as type 2 diabetes, cardiovascular disease, and fatty liver disease. While genome-wide association studies (GWAS) have been employed to identify genes associated with insulin resistance, the difficulty in measuring insulin resistance has led to limited sample sizes and reduced statistical power in previous GWAS studies. Conducting an expansive GWAS with a large sample size is essential to identify novel insulin resistance genes. The triglyceride to high-density lipoprotein ratio (TG/HDL) demonstrates a high correlation with insulin resistance phenotypes and is available in large cohorts, making it a potential surrogate marker for insulin resistance. Performing a GWAS of TG/HDL on 382,129 individuals in the UK Biobank, we identified 251 genomic risk loci, among which 62 displayed heightened significance compared to individual markers (TG or HDL), suggesting them as potential insulin resistance loci. We developed a systematic approach for the comprehensive analysis of these loci, prioritizing those that remain undiscovered and exhibit differential associations across different sexes. Through this analysis, we highlighted TNFAIP8, a gene involved in the regulation of apoptosis and immune responses, as a novel sex-dimorphic factor influencing insulin resistance. Our findings contribute to the genetic understanding of insulin resistance and identify potential sex-specific genes as therapeutic targets.

Skin specializes in a regional-specific manner, including the formation of distinct types of epidermally-derived glands during development, such as sweat glands in palmoplantar skin, sebaceous glands in hair-bearing skin, and meibomian glands in the eyelid. Our goal is to understand this regional specification. Palmoplantar skin is structurally and functionally unique, but transcriptional programs driving this specialization are unclear. Here, we use bulk and single-cell RNA sequencing of human palm, sole, and hip skin to describe the distinguishing characteristics of palmoplantar skin while also uncovering differences between palmar and plantar sites. Our approach reveals an altered immune environment in palmoplantar skin, with downregulation of diverse immunological processes and decreased immune populations. Further, we identify specific fibroblast populations that appear to orchestrate key differences in cell-cell communication in palm, sole, and hip. Dedicated keratinocyte analysis highlights major differences in basal cell fraction across three sites and demonstrates the existence of two spinous keratinocyte populations constituting parallel, site-selective epidermal differentiation trajectories. In summary, this deep characterization of highly adapted palmoplantar skin contributes key insights into the fundamental biology of human skin and provides a valuable data resource.In the eyelid, meibomian glands have received interest based on the association of observable gland dysfunction with dry eye disease. In particular, gland atrophy and changes in lipid quality are recognized as causing abnormal tear film break-up, leading to hyperosmolarity of the ocular surface, corneal epithelial damage and ocular surface inflammation. Here, we use single-cell and spatial transcriptomic approaches to characterize and spatially map the murine eyelid, with a particular focus on the meibomian gland. We identify sequential activation of lipid synthesis factors in meibocyte differentiation at the single cell level which we then validate spatially. Next, using an Awat2 KO mouse model, we characterize transcriptomic alterations, including an increase in immune populations and activation of psoriasis-like inflammatory pathways during meibomian gland dysfunction. Within the meibomian gland, we find marked expansion of the ductal epithelia and a shift in the meibocyte differentiation program. Finally, we identify a heightened immune environment in the aging eyelid, which is consistent both in healthy and KO mice.

Emerging genomic technologies that measure spatial information about RNA molecules promise to shed light on cell biology and function. However, most analytical techniques have primarily concentrated on spatial relationships at the multicellular and cellular scale without fully tapping into single-molecule spatial information. To address this gap, I introduce Bento, a toolkit designed for discerning spatial relationships at the subcellular scale. Bento incorporates a suite of statistical and machine learning methods within an intuitive Python programming interface, emphasizing the FAIR data management principles. To showcase its capabilities, I utilized Bento to study RNA localization changes in doxorubicin-treated cardiomyocytes profiled with spatial transcriptomics. Our findings reveal that doxorubicin-induced stress leads to the depletion of disease-associated genes in the endoplasmic reticulum, along with expression changes previously associated with doxorubicin-induced cardiotoxicity. This places the endoplasmic reticulum as a pivotal subcellular structure in the response to doxorubicin treatment. In essence, Bento emerges as a potent toolkit for the subcellular analysis of spatial transcriptomics data, paving the way for the discovery of new spatial relationships between subcellular structures and molecules. Furthermore, I have created a framework tailored to streamline image processing for spatial transcriptomics data called spotfish. Similar to Bento's ethos, spotfish is built in alignment with FAIR principles and leverages open-source standards like the Nextflow workflow language and Open Microscopy Environment file formats. Collectively, Bento and spotfish empower researchers to harness spatial transcriptomics technologies, enabling more comprehensive exploration of the spatial and molecular organization of cells at an unprecedented throughput.

CRISPR-Cas13 has been developed as a tool for genome editing by enabling precise RNA manipulation without permanently altering the genome. While its potential in therapeutics and diagnostics is promising, Cas13d’s off-target activities have limited its further use. To enhance our understanding of Cas13d’s binding specificity, we performed a massively parallel filter-binding assay probing the impact of mismatches in the target RNA on binding. Our results revealed that Cas13d binding is greatly affected by the specific position of the mutations in the target sequence, with less dependence on the type of mutation. Mismatches in the 5’ region of the target sequence were well tolerated, while those in the central region abrogated binding. These results emphasize the requirement for a complementary sequence in the central region for specific Cas13d binding. Additionally, this study expands the filter-binding method as a valuable tool for analyzing the binding specificity and off-target effects of Cas13d. Thus, our work lays a solid foundation for future high-throughput profiling of Cas13d binding and applying Cas13d for safe and effective transcriptional perturbation.

The recent development of single-cell RNA sequencing (scRNA-seq) technologies has revolutionized transcriptomic studies by revealing the genome-wide gene expression levels within individual cells. In contrast to bulk RNA sequencing, scRNA-seq technology captures cell-specific transcriptome landscapes, which can reveal crucial information about cell-to-cell heterogeneity across different tissues, organs, and systems and enable the discovery of novel cell types and new transient cell states. According to search results from PubMed, from 2009-2023, over 5,000 published studies have generated datasets using this technology. Such large volumes of data call for high-quality statistical methods for their analysis. In the three projects of this dissertation, I have explored and developed statistical methods to model the marginal and joint gene expression distributions and determine the latent structure type for scRNA-seq data. In all three projects, synthetic data simulation plays a crucial role.My first project focuses on the exploration of the Beta-Poisson hierarchical model for the marginal gene expression distribution of scRNA-seq data. This model is a simplified mechanistic model with biological interpretations. Through data simulation, I demonstrate three typical behaviors of this model under different parameter combinations, one of which can be interpreted as one source of the sparsity and zero inflation that is often observed in scRNA-seq datasets. Further, I discuss parameter estimation methods of this model and its other applications in the analysis of scRNA-seq data.My second project focuses on the development of a statistical simulator, scDesign2, to generate realistic synthetic scRNA-seq data. Although dozens of simulators have been developed before, they lack the capacity to simultaneously achieve the following three goals: preserving genes, capturing gene correlations, and generating any number of cells with varying sequencing depths. To fill in this gap, scDesign2 is developed as a transparent simulator that achieves all three goals and generates high-fidelity synthetic data for multiple scRNA-seq protocols and other single-cell gene expression count-based technologies. Compared with existing simulators, scDesign2 is advantageous in its transparent use of probabilistic models and is unique in its ability to capture gene correlations via copula. We verify that scDesign2 generates more realistic synthetic data for four scRNA-seq protocols (10x Genomics, CEL-Seq2, Fluidigm C1, and Smart-Seq2) and two single-cell spatial transcriptomics protocols (MERFISH and pciSeq) than existing simulators do. Under two typical computational tasks, cell clustering and rare cell type detection, we demonstrate that scDesign2 provides informative guidance on deciding the optimal sequencing depth and cell number in single-cell RNA-seq experimental design, and that scDesign2 can effectively benchmark computational methods under varying sequencing depths and cell numbers. With these advantages, scDesign2 is a powerful tool for single-cell researchers to design experiments, develop computational methods, and choose appropriate methods for specific data analysis needs.My third project focuses on deciding latent structure types for scRNA-seq datasets. Clustering and trajectory inference are two important data analysis tasks that can be performed for scRNA-seq datasets and will lead to different interpretations. However, as of now, there is no principled way to tell which one of these two types of analysis results is more suitable to describe a given dataset. In this project, we propose two computational approaches that aim to distinguish cluster-type vs. trajectory-type scRNA-seq datasets. The first approach is based on building a classifier using eigenvalue features of the gene expression covariance matrix, drawing inspiration from random matrix theory (RMT). The second approach is based on comparing the similarity of real data and simulated data generated by assuming the cell latent structure as clusters or a trajectory. While both approaches have limitations, we show that the second approach gives more promising results and has room for further improvements.

Aging is a risk factor for many diseases, and understanding the hallmarks of aging is crucial for developing interventions to promote healthy aging. Despite the conservation of the aging hallmarks across phylogeny, the order in which these hallmarks appear and their relationship to each other remains unclear. To address these questions, we used the budding yeast Saccharomyces cerevisiae as a model organism due to its short lifespan and genetic tractability. The Li lab has developed a high-throughput microfluidics device to trap and monitor individual yeast cells over time. Using cell tracking technology, we quantified tagged proteins and classified different organelle phenotypes to predict lifespan based on cellular characteristics such as protein levels and organelle morphology. Our analysis revealed that the proteasome marker Rpn11p and size scaling is a strong predictor for proteostasis stress and mitochondrial dysfunction hallmarks. We identified slope changes and areas of certain organelles as strong predictors of replicative lifespan with differences in short-lived versus long-lived cells in maintaining a stable scaling ratio. We investigated the relationship between organelle scaling and protein dilution and found certain nuclear proteins to decrease with the increasing nuclear size. Our analysis revealed the proteasome as a regulator of the scaling ratio and influencer of the mitochondrial morphology and activity. We provide new insights into the mechanisms underlying aging and potential targets for interventions to promote healthy aging.

iPSC reprogramming is a process that converts somatic cells, such as fibroblasts, into induced pluripotent stem cells (iPSCs). iPSCs exhibit the properties of embryonic stem cells, including the ability to give rise to every cell type in an organism. iPSCs are of prime interest in the field of regenerative medicine because they can be used to generate a wide range of cells, which can potentially replace damaged cells in a patient without the threat of immune rejection. Reprogramming involves dynamic changes in chromatin accessibility, which are critical for the conversion of fibroblast cells to iPSCs. Recent technological advances in sequencing and machine learning can be applied to further explore the mechanisms by which cells are able to silence somatic genes and activate pluripotent gene networks. The following chapters investigate the mechanism of cellular reprogramming and the practical applications of stem cell properties. In Chapter 1, a meta-analysis of comparable epigenomic and transcriptomic human and murine data sets was conducted to identify consistent changes in chromatin accessibility regions during reprogramming. Integrative approaches were then used to construct stage-specific regulatory networks of human cellular reprogramming from gene expression and chromatin accessibility data. In Chapter 2, stem cells were utilized for disease modeling and pharmaceutical testing. In Chapter 3, artificial intelligence methods were employed to aid in discovery of potential pharmaceuticals for preclinical, in vitro testing. The primary focus of regenerative medicine is to efficiently derive human stem cells in order to treat a host of diseases and disabilities. Overall, this dissertation provides a significant contribution to understanding the process by which cells become pluripotent, which will aid in surmounting current challenges associated with stem cell technology and potentially lead to revolutionary therapeutics. In attempt to help actualize this, this dissertation also heavily investigates the practical applications of stem cells and AI-driven drug discovery as avenues to propose, develop and test medical solutions.

Basal cell carcinoma (BCC) is the most common cancer in the US, but treatment of advanced tumors becomes complicated by drug resistance. Current targeted therapies for advanced BCC suppress inappropriate activation of the Hedgehog signaling pathway and result in potent tumor regression. Yet, they are only effective in 40% of the tumors, with 20% of the initial responders acquiring resistance to these drugs within one year. Therefore, there is an urgent need for alternative methods to treat advanced BCCs and possibly halt tumor development before they reach later stages. This thesis work investigates different mechanisms of modulating BCC tumor growth and progression to understand the mechanisms in the skin environment that dictate whether these tumors will form or regress. Understanding what dictates tumor progression versus intrinsic tumor control holds the potential to enable the development of novel and combination treatment strategies for BCC patients. We used a transgenic BCC mouse model that grows microtumors that spontaneously regress over time to study the role of immune system interactions in promoting tumor regression in BCC. The study aims to define the immune response against BCC tumors and determine the regulatory role of effector lymphocytes in regression. We use this inducible transgenic model coupled with varying degrees of immune cell perturbations, including genetic knockout of lymphocytes and antibody depletions of cell subtypes, to determine the effect on tumor burden. Single-cell RNA sequencing and immunofluorescence microscopy reveal an upregulated innate immunity cytokine response and a T cell Th1-driven response. The CellChat algorithm for inference of communication probability in single-cell data reveals IFNγ and CCL5 signaling from T cells to BCC cells, neutrophils, macrophages, or other T cells. There is also evidence of immune regulation by lymphocytes in the BCC tumor microenvironment, as lack of αβ T cells and adaptive immunity impairs efficient tumor growth (p

During vertebrate embryogenesis, tissue morphology and cell differentiation are constantly influenced by, and responding to, microenvironment cues such as physical forces and associated biochemical signaling processes. Few tissues exemplify the complex interplay of these coexisting processes better than tendons, the extracellular matrix (ECM) rich connective tissues which attach muscle to bone, cartilage, and soft tissues, and coordinate the optimal transfer of force from of muscle contraction to the skeleton. Though in recent years, much has been elucidated about gene regulatory networks coordinating tendon tissue morphogenesis and tendon fibroblast (tenocyte) fate specification/differentiation, very little is known about how heterogenous tenocyte populations sense and respond to muscle contraction forces and uniquely modify their ECM organization and composition in vivo. Current knowledge of tenocyte mechanotransduction has been primarily informed by studies in adult in vivo models, and generally in tendons restricted to mammalian limbs. Zebrafish embryonic tendons provide a particularly useful model to address these research questions, as heterogenous tenocyte populations inhabiting different tendons can easily be studied across developmental timepoints. In this thesis work, I leveraged next generation bulk and single-cell sequencing approaches combined with in vivo functional perturbations and protein binding assays to gain a more defined, holistic understanding of tenocyte-ECM functional interactions during vertebrate development. I investigate tenocyte transcriptional responses to the onset of muscle contraction force during development by conducting RNA sequencing on sorted tenocytes from zebrafish embryos at developmental stages before and after the onset of muscle activity. I show that onset of muscle contractile forces leads to a specific transcriptional change in tenocytes, and using muscle-paralysis perturbation assays confirm that three novel tenocyte subpopulation markers, matn1, klf2a, and mxra5b are differentially expressed. Further, I show that variations in magnitude of muscle contractile force lead to unique transcriptional dynamics in specific tenocyte subpopulations in vivo, suggesting that tenocytes fine-tune ECM gene expression to adapt individual tendons to specific force conditions.In vivo, tendon-like structures connect muscles not only to bone and cartilage, but to soft tissue such as eye sclera, and transmit varying intensity of forces from muscle contraction. Recent sequencing studies have begun to unravel distinct tenocyte subpopulations but have primarily looked at only gene regulatory networks of tenocytes inhabiting individual tendons or tendon subregions, prompting the question as to whether tenocyte populations are transcriptionally distinct within and between distinct tendons. To understand this tenocyte transcriptional heterogeneity at the level of both inter- and intra-tendon tissues, I conducted single-cell RNA sequencing (scRNAseq) on sorted cranial tenocytes from dissected zebrafish heads. I show for the first time that cranial tenocyte populations transcriptionally cluster into both spatially distinct tendons and functionally distinct intra-tendon regions, and that this clustering is driven not only by generalized patterning genes but also by specific ECM components. I further show that individual intra-tendon attachment zones have distinct ECM transcriptional signatures between tendons, suggesting that a combination of developmental programming and varying levels of force at attachment zones underlies tenocyte heterogeneity. Next, I identify a previously unknown population of Wnt-signaling responsive tenocytes populating multiple tendons in the cranium and show through independent Wnt signaling perturbation assays that a proper balance of Wnt signaling is required for proper patterning of myotendinous junctions of the jaw. Through computational cell-communication inference software, I hypothesize that specific Wnt ligands and receptors are involved in this patterning process. These findings revealed a deeper layer of complexity to the understanding of tendon biology, demonstrating that transcriptional diversity of tenocytes is driven by the dual characteristics of specific attachment zone physiology and the load capacity of the tendon ECM.Lastly, I investigate how tenocyte fate specification occurs at the level of transcriptional control by identifying putative binding partners to key tenocyte fate determining transcription factors Scleraxis and Mohawk. Through biochemical protein binding assays and immunofluorescence I show that Scleraxis binds a family of Class II helix-loop-helix transcription factor E-box proteins TCF3, TCF4, and TCF 12, and that these proteins are individually sufficient to drive Scleraxis translocation from the cytoplasm to the nucleus. This reveals a novel mechanism for initiation of Scleraxis transcription, which has sweeping implications for further understanding of tenocyte fate determination and maintenance. In this thesis, I have provided unique insights into novel roles for muscle-contraction force on tendon tissue remodeling, tenocyte transcriptional heterogeneity, and lineage specification.

The pervasive presence of guanine quadruplex (G4) structures within regulatory regions of the genome has spurred intense research efforts to understand their roles in modulating cellular pathways. Numerous investigations affirm that both DNA and RNA G4s actively partake in pivotal biological processes, including DNA replication, transcription, RNA metabolism, translation, and telomere maintenance. However, further investigation is needed to understand the molecular mechanisms through which these secondary nucleic acid structures modulate biological processes. This dissertation focuses on utilizing bioinformatic analyses of publicly accessible datasets to unravel multifaceted roles of G4 structures in histone modifications, three-dimensional chromatin configurations, RNA metabolism, and telomere maintenance.In chapter 2, we revealed the G4 co-localization pattern with transcription factors and constructed an interaction network of candidate G4-interacting proteins. Moreover, we explored the interplay between G4 structures and histone marks, unveiling G4 structures as active transcription marks and potential regulators for histone modifications.In chapter 3, we conducted an intuitive overlapping analysis of previously published RNAPII ChIA-PET and BG4 ChIP-seq data, and our work revealed a strong positive correlation between RNAPII-linked DNA loops and G4 structures in chromatin. In conjunction with HiChIP-seq and RNA-seq, we unveiled vital role of DNA G4 in RNAPII-associated DNA looping and transcription regulation.In chapter 4, we employed a bioinformatic approach based on the analysis of overlap between RNA G4 (rG4)-seq analysis and eCLIP-seq datasets generated from the ENCODE project. We identified a large number of candidate rG4-binding proteins. We validated that one of these proteins, G3BP1, is a direct binder of rG4 structures, and documented a rG4-dependent function in regulating mRNA stabilities and translation efficiencies.In chapter 5, I proposed a novel approach for identifying putative telomere-binding proteins through enrichment analysis of ChIP-seq datasets covering zinc finger proteins. Three prominent targets, i.e., ZNF24, ZNF316 and ZBTB33, exhibited significant enrichment with telomeric sequences. A detailed examination of ZBTB33 suggested a potential G4-dependent telomere binding activity of the protein.In summary, this dissertation introduces a pioneering bioinformatic approach for investigating the intricate interplay between G4 structures and other cellular regulatory mechanisms, identifying new RNA-binding proteins and probing potential G4-dependent telomere-binding proteins. These insights underscore the regulatory significance of G4 structures and shed light on their intricate roles in the nuanced modulation of cellular processes.

Heterosis, or hybrid vigor, refers to the phenomenon in which F1 offspring of two genetically distinct parental lines exhibit phenotypes outside the range of its parents. Agricultural breeding programs commonly utilize heterosis to enhance yield, disease resistance, and other desirable traits in both crops and livestock, despite the labor-intensive field tests required for hybrid breeding programs. Despite its importance, the mechanisms by which genetic variation between two parents combine to produce non-additive phenotypes in hybrids remain unclear. To identify molecular components that may contribute to heterosis, we analyzed paired proteomic and transcriptomic data from leaf tissue of maize hybrids and their inbred parents. Expression levels of plastid proteins involved in translation and photosynthesis were increased in seedling leaf tissue of hybrids relative to mid-parent and were positively correlated with heterosis levels in adult plants. Conversely, levels of proteins involved in stress responses were reduced in seedling leaf tissue of hybrids relative to mid-parent and were negatively correlated with heterosis levels in adult plants. An ethylene biosynthesis mutant copied the hybrid proteome, indicating that the most of the altered protein levels in hybrids are downstream of the reduction in ethylene biosynthesis. Protein expression patterns across the developmental gradient of hybrid leaves were altered in comparison to the inbred parents. Proteins involved in photosynthesis and translation were reduced in the basal zone of the leaf but increased in the mature zone. This suggests that hybrids have more precisely controlled expression patterns across the developmental gradient which could contribute to their greater fitness. The same expression patterns were observed in the ethylene mutant, indicating that reduced ethylene biosynthesis largely mediates the developmental differences between hybrids and inbreds.

Understanding the transcriptional consequences of oncogenic mutations is an important goal that may reveal new therapeutic targets for diverse cancers. Although single-cell methods hold promise for this task, it remains non-trivial to isolate and sequence DNA and RNA from the same cell at scale. Here we present a statistically motivated strategy that utilizes multiscale and multiomic analysis of individual human tumor specimens to deconstruct intra-tumoral heterogeneity by clarifying clonal populations of malignant cells and their transcriptional profiles. By combining deep, multiscale sampling of IDH-mutant astrocytomas with integrative, multiomic analysis, we reconstruct and validate the phylogenies, spatial distributions, and transcriptional profiles of distinct malignant clones. We identify a core set of genes that is consistently expressed by the truncal clone, including AKR1C3, whose expression is associated with poor outcomes in several types of cancer. Some derived clones exhibit significant enrichment with gene sets representing glioblastoma subtypes and nonmalignant cell types, including ependymal cells. Importantly, by genotyping nuclei for truncal mutations, we show that existing strategies for inferring malignancy from gene expression profiles of single cells may be inaccurate. Furthermore, we find that transcriptional phenotypes of malignancy persist despite loss of the mutant IDH1 protein following chr2q deletion in a subset of malignant cells. In summary, our study provides a generalizable strategy for precisely deconstructing intra-tumoral heterogeneity and clarifying the molecular profiles of malignant clones in any kind of solid tumor.We extend this approach to a metaanalysis of the cell-type specific dysregulation in the glioma microenvironment. Using the same statistically motivated approach to take advantage of inherent patterns of cell-type specific coexpression, we characterize the differentially expressed cell-type specific transcriptome. We perform this process on thousands of samples and hundreds of datasets from both glioma and normal samples. Finally by taking the difference in the in silico derived differential expression metrics for each cell type in glioma and normal contexts, we identify ideal markers of each cell type specifically in glioma but not normal and validate and filter them using orthogonal datasets. In summary, our study provides a generalizable strategy for precise identification of cell-type specific dysregulated genes using abundant bulk transcriptome data for any disease state involving solid tissues.

Microbial communities are hard to study, yet their importance is hard to overstate. They inhabit every surface on Earth and are core participants in global biogeochemical cycling, but this is under threat due to anthropogenic climate change.One important microbiome is those communities that can be found decomposing leaf litter, the top layer of soil. This decomposition is critical for recycling nutrients back into the soil for the next generation of life.Here, I made use of microbial “cages” that permit water and nutrients to flow in and out while preventing microbial dispersal, allowing the study of microbes in situ. With the help of amplicon, metagenomic, and whole genome sequencing, and from the perspectives of ecology, evolution, and physiology, I dove into a single bacterial genus, Curtobacterium, to investigate the amount of ecologically important variation from closely related strains.I found evidence of variation between clades, with some strains surviving 10x better under desiccation than others. Further, variation exists between subclades, that might be missed using well-established methods of community analysis. I also investigated within a subclade, finding evidence that evolutionary processes operate during the timescale of ecological processes.

There are numerous signaling dynamics involved in gastrulation. Using biophysical modelling, we find that liquid-liquid phase separation significantly improves the processing of transcriptional factor β-catenin in the canonical WNT signaling cascade. Using a finite element method, we find that a nucleating region can facilitate a nearly two-fold increase in β-catenin processing. Moving away from microscale biophysical dynamics, we employed a large-scale loss of function drug screen on 2D cultured embryo bodies and found that the 2D system can be used as a powerful, cheap, and reproducible teratogen screening platform and discovered the surprising role of inflammatory and immunological signaling on human gastrulation.

The placenta is a temporary organ that develops during pregnancy. Prior work has found that placentas are genetically heterogeneous due to developmental bottlenecks in trophoblast differentiation. The conflicting relationship between confined placental mosaicism and adverse pregnancy outcomes is hypothesized to be from nonuniform mutations throughout the placenta resulting from mosaicism. Retrospective lineage tracing, particularly in trophoblasts, could be key to understanding how derivations of cell fate connect to the development of placental dysfunction. As such, it is important to characterize the relationship between somatic genetic variation, cell types, and lineage tracing in both healthy and diseased placentas. We identified rare, damaging, loss-of-function mutations in placentas with pre-eclampsia and other histopathological lesions that may be clinically relevant. We developed a computational pipeline to perform single cell/nuclei variant calling using STARSolo and GATK Best Practices Workflow for RNA-seq Short Variant Discovery. We also employed VarTrix to perform single-cell genotyping. Using both methodologies, we generated single-nucleotide variant by cell barcode matrices for four 10X Multiome single-nuclei RNA-sequencing samples obtained from different sampling sites in a one human term placenta. Lastly, we performed trajectory inference to determine the relationship between mutational burden and lineage in trophoblasts and find that mutational burden is able to differentiate between different syncytiotrophoblast clusters.