106261 research outputs found
Sort by
NEW CONNECTIONS IN THE GENE REGULATORY NETWORK UNDERLYING CELL FATE SPECIFICATION IN C. ELEGANS POSTEMBRYONIC MESODERM DEVELOPMENT
No full text194 pagesCell fate specification during development requires the combinatorial actions of transcription factors and cell-cell signaling. The C. elegans postembryonic mesoderm, or M lineage, is derived from a single multipotent precursor cell, which during hermaphrodite postembryonic development produces 32 cells of six different types. The M lineage therefore provides an excellent system to dissect the regulatory mechanisms underlying fate specification. My thesis work centers around a key transcription factor, SEM-2, which belongs to the SoxC group, and its roles in M lineage development. By analyzing animals carrying a partial loss-of-function mutation in SEM-2, I uncovered previously unappreciated functions and interactions of SEM-2 in M lineage development. First, in addition to its role in specifying a ventral M lineage fate, the sex myoblast (SM) fate, SEM-2 also functions in the dorsal M lineage, where it antagonizes the expression and function of the forkhead transcription factor LET-381/FoxF/C. Second, SEM-2 is not only required for specifying the SM fate, but it is also essential for the proliferation and diversification of the SM lineage, particularly the differentiation of type II vulval muscles required for egg-laying. Third, SEM-2 appears to directly regulate the expression of hlh-8, which encodes a basic helix-loop-helix Twist transcription factor that plays critical roles in the proper patterning of the M lineage. My findings suggest that the SoxC-Twist axis, including the downstream targets of Twist, represents an evolutionarily conserved regulatory cassette important in metazoan development. There are three SoxC proteins in mammals, Sox4, Sox11 and Sox12. Mutations in Sox4 and Sox11 are associated with a neurodevelopmental disorder called Coffin-Siris syndrome (CSS). Many CSS-associated mutations affect conserved residues in SoxC proteins. I introduced a SoxC CSS-associated mutation into sem-2 in worms and confirmed that it is a partial loss-of-function mutation that likely causes defects in humans due to haploinsufficiency. Further phenotypic analysis of the mutant worms revealed that SEM-2 not only functions in the M lineage, but also in other mesodermal tissues, specifically in the defecation muscles, possibly by acting through hlh-8/CeTwist. Altogether, my work identified new interactions in the gene regulatory network underlying C. elegans postembryonic development and adds to the general understanding of the structure-function relationship of SoxC proteins.2027-06-1
THE VAPOR-SOLVENT EFFECT ON INITIATED CHEMICAL VAPOR DEPOSITION (ICVD) POLYMERIC THIN FILMS
No full text29 pagesInitiated Chemical Vapor Deposition (iCVD) is a solvent-free method that integrates polymerization and polymer processing into a single step to produce polymer thin films that are uniform and defect-free in benign synthesis conditions that retain functional moieties. Recently, the use of vapor solvents during iCVD has emerged as a promising strategy to enhance deposition rates and control film properties. This work aims to expand the understanding of iCVD films influenced by vapor solvents, with a focus on poly(acrylic acid) (PAA) thin films fabricated using triethylamine (TEA) as the vapor solvent. The research investigates the effects of vapor solvent inclusion on thermal, chemical, and physical properties, including glass transition temperature (Tg), polymer composition, and molecular weight. Key findings include the incorporation of TEA vapor solvent in the polymer films post-deposition and that vapor solvent can be removed from the film by heating above its boiling point, mitigating concerns about vapor solvent retention during applications. Other findings include depressed Tg values and increased molecular weight for vapor-solvent-assisted PAA films. These results provide critical insights into the use of vapor solvents in iCVD, highlighting benefits such as increased deposition rates, higher molecular weights, and the non-intrusive nature of vapor solvents upon removal
EXTRAPOLATION AND BUBBLES: MORE REALISTIC ASSUMPTIONS
No full text36 pagesWhile the Barberis et. al (2018) model effectively captures the empirical featuresof overvaluation and high trading volume during bubbles, some empirical phenomena remain unaddressed. For example, the original model requires a shortsale constraint to generate bubbles, as fundamental traders would otherwise short too aggressively against overpricing, effectively removing the bubble. In addition, the model generates three peaks in trading volumes during a bubble which is not in accordance with the real data. The model can be made more realistic along a number of dimensions: we can allow for heterogeneous sentiment; for fundamental traders perceiving a slow correction of mispricing; and for two components to the sentiment process that drives investor demand
Scalar Curvature and its Topological Implications
No full text156 pagesScalar curvature and its implications on topology have been a focus of intense study in recent years. In this thesis, we compute Yamabe invariants of certain compact manifolds, including in all dimensions. We provide a new computation of the Yamabe invariant of via harmonic functions and a monotonicity formula along the level set of the harmonic functions; this is a joint work with Liam Mazurowski. In collaboration with Dongyeong Ko, we establish a scalar curvature comparison and rigidity theorem in dimension via capillary minimal hypersurfaces in compact manifolds with non-empty boundary. We also formulate a conjecture concerning -intermediate Ricci curvature that interpolates between the Cheeger–Gromoll splitting theorem and the conjecture, highlighting a new perspective on the interplay between Ricci and scalar curvature. We prove the conjecture in dimensions and verify most cases in dimension . As a corollary, we prove that -dimensional aspherical manifolds does not admit a metric with positive -intermediate curvature, which is only slightly stronger than scalar curvature, providing evidence on the conjecture in dimension ; this is a joint work with Liam Mazurowski and Tongrui Wang
The Terraqueous Romantic: Nineteenth-Century Women Writers Between Land and Sea
No full text284 pagesScholarly emphasis on the sea as a site of masculinizing labor and adventure has failed to recognize how women harnessed the liberatory affordances of maritime circulations for themselves. This dissertation addresses that oversight, analyzing how nineteenth-century North American women writers engaged oceanic histories of resistance to colonialism, ecological destruction, and racialized oppression. Focusing on the work of Nancy Prince, Frances Ellen Watkins Harper, Emily Dickinson, and E. Pauline Johnson, this project considers the waterborne conflicts and intimacies that formed along the United States’ northern border as well as its seacoasts. It moves from circum-Atlantic contexts to the Great Lakes—the so-called “sweetwater seas”—to establish women writers’ interest in a language of maritime trade, travel, and resistance, revealing how oceanic imaginaries circulated from ship and port into the home. These writers’ preoccupation with liminal—and littoral—zones of oceanic encounter constitutes an ecopolitical subjectivity I call the “terraqueous romantic.” This perspective controverts arguments that romanticism neglects the ecological complexity of the ocean in favor of an “anthropocentric” or “egotistical” sublime. The women I analyze do not turn to the sea to assert their mastery over nature but to confront the power of imperialism and environmental exploitation; their interest in oceanic metaphor, metonymy, and apostrophe suggests the sea’s erosive, destabilizing force in nineteenth-century discourses of race, gender, and nation. By identifying the ecocritical implications of this gendered preoccupation the sea, this dissertation more adequately historicizes theories of human-ocean relation, revealing the ongoing presence of women in oceanic contexts and literary histories from which they have long been excluded.2027-06-1
LEVERAGING COMPUTATIONAL TOOLS IN PRECISION MEDICINE: INVESTIGATING PROTEIN INTERACTION NETWORKS TO UNCOVER DISEASE PERTURBATIONS, FUNCTIONAL IMPACTS AND THERAPEUTIC STRATEGIES FOR TARGETED TREATMENT DEVELOPMENT
No full text182 pagesThe human proteome comprises proteins that rarely act alone, often carrying out their functions through direct interactions with other proteins or as part of larger complexes. Perturbations in these proteins can destabilize their structure, hindering their ability to form these protein-protein interactions (PPIs). These disrupted PPIs impact biological processes and can contribute to the development of various diseases, posing a significant challenge in treating conditions with complex, multifactorial pathophysiology, such as Alzheimer’s disease (AD) and cancer. Thus, achieving a comprehensive understanding of the molecular mechanisms driving these complex conditions becomes crucial for developing more targeted and effective therapeutics. This dissertation utilizes computational tools to analyze protein interaction networks, identify disease-related perturbation and deepen our understanding of therapeutic mechanisms, with the ultimate aim of advancing the development of more targeted treatments. Chapter 1 introduces relevant background to set the context for the work presented herein. Chapter 2 provides a comprehensive review of the advancements in both the experimental and computational aspects of cross-linking mass spectrometry (XL-MS) workflow. XL-MS is a powerful technique for elucidating PPIs and their structural context that, when integrated with multi-omics data, can be utilized by machine learning algorithms to predict how disruptions in protein interactions affect cellular networks and drive disease progression. Chapter 3 describes my contributions to development of the PIONEER web server tool and on-demand prediction pipeline, associated with the deep learning framework, which uses available structural information to make predictions about PPI interfaces. Such web server tools offer a system-level perspective on perturbation in PPI interfaces and their downstream effects. By offering a user-friendly web-based tool with predictive capabilities that integrates various levels of protein structure information with disease-associated mutations, PIONEER represents a notable step forward in building computational tools accessible to the wider scientific community to advance precision medicine. In Chapter 4, computational approaches are applied to predict mechanisms of pCDP-DB, a novel cis-locked cyclic dipeptide that alters the fate of the amyloid precursor protein (APP) and its cleavage products, thereby reducing amyloidogenic processing of APP and/or inducing clearance of sAPP_ and A_ from H4 neuroglioma cells. Previously obtained pCDP-DB interactomes were analyzed by generating PPI networks, applying a clustering algorithm to elucidate highly interconnected subgroups of proteins, then identifying statistically enriched pathways within these communities. Our findings reveal pleiotropic effects of pCDP-DB on several pathways dysregulated in AD. These include phago-lysosomal pathways and proteasome/autophagy systems that could contribute to enhanced A_ clearance. Integration of these findings with existing literature contextualizes the role of enriched pathways. This combined approach of data-driven discovery and literature-based validation generates predictions about pCDP-DB’s mechanisms of action, guiding further experimental studies for mechanistic validation. By systematically analyzing PPI networks, this work establishes a framework for uncovering protein communities, identifying dysregulated pathways, and developing more targeted treatments for human diseases. Overall, this research underscores the transformative potential of computational tools in precision medicine, bridging the gap between molecular understanding of complex diseases and the development of more precise therapeutic strategies
BETWEEN AFFECT AND EFFECT: COMPUTATIONAL EXPLORATION OF THE LATENT SPACE BETWEEN ORNAMENT AND STRUCTURE
No full text65 pagesArchitectural design has long explored ornament and its relationship to structure, sometimes treating it as essential, sometimes as a complement, and other times as decoration. This paper explores the use of computational design methodologies to examine this relationship, specifically investigating optimization frameworks to generate structurally performing yet aesthetically specific designs. The research investigates two methods for combining aesthetic and performance objectives through a case study of ornamental patterns found in mosques and churches. The first method takes a pixel-based approach, combining topology optimization with a diffusion-based image generation model. The second method uses an evolutionary optimization algorithm to adjust a parametric design space, where a Vision Transformer model was used as a surrogate to embed designer aesthetic preferences into the optimization process, and Finite Element Analysis (FEA) evaluated structural performance based on displacement and weight criteria. The parametric approach was applied across case studies, including block tile optimization, modular element optimization, and a material-constrained dome fabricated from plywood. The results show the potential of integrating Machine Learning (ML), FEA, and optimization to address the fuzzy nature of aesthetic features while promoting structural performance.2027-06-1
The Effect of Visible Enforcement of Bald and Golden Eagle Protection Act on Eagles' Relative Abundance
No full text87 pagesWind energy harms birds, constituting violations of the mandatory Bald and Golden Eagle Protection Act. This Act was not visibly enforced until late 2013. We examine the effectiveness of this visible enforcement on eagles’ abundance protection through the pathway of wind energy in the continental United States. We find that bald eagles’ abundance continues to decline after 2013 but at a reducing rate. Golden eagles’ abundance increases immediately right after the enforcement but remains stable in later periods. We also discover that wind turbine presence negatively affects bald eagles before the enforcement, but they are protected in turbine-presence counties after 2013 since this enforcement effectively regulates wind energy development. Wind turbine presence has no significant effects on golden eagles’ abundance both before and after the enforcement. There are also no significant changes in how the number of wind turbines and wind energy rated capacity affect both eagles’ abundance after 2013
FUNDAMENTAL PROCESSES IN INITIATED CHEMICAL VAPOR DEPOSITION
No full text247 pagesPolymer chemical vapor deposition (CVD) offers superior conformality and substrate compatibility compared to solution-based thin film fabrication. Unique among polymer CVD processes, initiated CVD (iCVD) decouples radical generating reactions from polymerization reactions, enabling use of mild conditions, and diverse polymer chemistries. Most iCVD research falls into one of three categories: (i) implementation, (ii) process innovations, and (iii) post-deposition modifications. This dissertation targets critical knowledge gaps in these areas by investigating the fundamental processes relevant to them. Beginning with implementation, Chapter 2 addresses the gap in understanding of defects which might pose a risk to use of iCVD in industries requiring smooth, uniform films, such as optical coatings and soft electronics. Films fabricated using iCVD at sample temperatures below ~30oC are shown to be susceptible to defects originating from build-up of oligomeric species (n < 10) and prevention methods are developed. Chapters 3 and 4 address the understudied iCVD vapor phase. Even though innovations like batch iCVD and condensed droplet polymerization were built on understanding of the vapor phase, no models of the iCVD vapor phase have been validated experimentally. Chapter 3 first investigates heat transfer and mixing in the iCVD reactor, introducing the Peclet and Knudsen numbers to quantify these effects. Chapter 4 then thoroughly characterizes the reactive vapor phase, developing a robust model of initiator (di-tert-butyl peroxide) decomposition at a range of flowrates, pressures, temperatures and monomer (cyclohexyl methacrylate) compositions. Moving to post-deposition modification, Chapter 5 investigates post-deposition base catalyzed hydrolysis of two fluoroacrylic polymers. The high pH conditions required to hydrolyze the passivating fluoroacrylate moiety pushes the limits of post-deposition modification. Quantification of hydrolysis kinetics shows that both crosslinking and covalently grafting the film to the substrate are necessary. The eventual degradation of the grafting agent indicates a need for improved grafting procedures
Towards Accurate and Scalable Performance Modeling and Benchmarking of Cloud Applications
No full text165 pagesCloud computing now powers a significant portion of global computation, supporting everything from latency-sensitive interactive services to artificial intelligence (AI) models. As these applications continue to shape the future of cloud infrastructures, understanding their behavior is critical. In this context, accurate and scalable performance modeling and benchmarking have become essential for optimizing system performance and guiding the development of cloud infrastructure. Agile, efficient, and precise modeling and benchmarking tools can provide invaluable insights for fleet design and optimizing efficiency. Within a cloud provider, these tools can support hardware and system optimization (e.g., GPU or ASIC accelerator design), performance characterization and analysis, design space exploration, and bug reproduction. Furthermore, they can be shared with external hardware vendors for early-stage performance testing, evaluation, and joint hardware/software co-design, all while requiring minimal infrastructure support and enabling a streamlined IP-sharing framework. The highly diverse and rapidly evolving landscape of cloud applications presents significant challenges in developing modeling and benchmarking tools that accurately capture performance characteristics. For instance, while open-source benchmarks have seen significant advancements, their update speed, application diversity, and complexity cannot keep pace with the constantly changing and varied applications in real cloud deployments. As a result, engineers and researchers have to manually adapt existing production or open-source workloads into forms suitable for benchmarking. This process requires substantial expertise and a deep understanding of the workloads, making it a non-trivial investment. Consequently, maintaining and updating these benchmarks to keep up with the fast pace of cloud application development incurs high costs. Therefore, there is a strong need for new methodologies that enable efficient and accurate modeling and benchmarking of cloud applications. This thesis presents novel solutions to address these challenges. We first introduce Ditto, an automated cloning framework for end-to-end interactive cloud services, including both monolithic applications and microservices. It begins by capturing the dependency graph across services using distributed tracing, then reconstructs the high-level control and data flow within each service. Finally, Ditto generates system calls and user-level assembly to capture both on-CPU and off-CPU behavior. This process is fully automated, allowing users to clone and benchmark services without needing expertise in their implementation. Our evaluation demonstrates that synthetic applications generated by Ditto respond to changes in input load, platform, resource allocation, and deployment configuration in the same way as the original workloads. Next, we present Mystique, an efficient and scalable framework for generating AI benchmarks. By leveraging execution traces directly captured from production workflows, Mystique generates benchmarks using a "replay-as-benchmark" approach. The trace records runtime information of a model at the operator level, and Mystique faithfully replays it to accurately reproduce the original performance. We demonstrate that our methodology generates AI benchmarks that closely mirror the original applications, both in terms of execution time and system-level metrics, while remaining easy to use and portable across platforms without the need for regeneration. We also highlight several use cases for Mystique, including early-stage platform evaluation, subtrace replay, and scaled-down performance testing. Finally, we discuss Lumos, a trace-driven performance modeling and estimation toolkit for large-scale training of large language models (LLMs). By leveraging built-in profiling tools from machine learning (ML) frameworks, Lumos constructs a comprehensive execution graph to capture the runtime behaviors of LLMs and build accurate performance models. It also provides users with a convenient way to explore various model and deployment configurations through graph manipulation and simulation, streamlining the exploration process. We evaluate Lumos using various GPT-3 model variants on a production-scale cluster and demonstrate that it accurately reproduces and predicts execution times and detailed performance characteristics across different models and configurations