80264 research outputs found
Sort by
Development of experimental platforms for ultra-high- throughput exploration of complex genetic design spaces
No full textCells sense and process signals from their environment to execute a diverse array of tasks, ranging from proliferation and differentiation to programmed cell death. Inspired by the capabilities of biological systems to carry out sophisticated computations, the field of synthetic biology aims to use nucleic acid-encoded “synthetic” regulatory programs to quantitatively engineer novel cellular behaviors for environmental, biotechnological, and therapeutic purposes. Like many forms of engineering, synthetic biology projects follow a design-build-test-learn cycle: an iterative process of constructing, assaying, and modifying genetic circuits to achieve desired phenotypes. However, unlike more established forms of engineering, we do not have a quantitative set of core principles that describe the complexities of all biological activity. This limitation is pronounced in mammalian synthetic biology, where lengthy design campaigns and an incomplete understanding of the system make precisely programming cellular functions difficult. One approach to addressing challenges in synthetic biology is to increase the pace and scale of data acquisition and allow experimental data to replace hypotheses as the cornerstone of decision-making. Here, I present a suite of molecular biology and cell engineering tools that lay the foundations of a novel platform designed to enable high-throughput construction and quantitative assessment of large and complex genetic design spaces. This platform, named CLASSIC (combinatorial large-scale assembly and short-range sequencing for investigating genetic complexity), offers a novel opportunity to generate genotype- to-phenotype (G2P) maps for hundreds of thousands of multi-kilobase genetic circuits in a single experiment. We show proof-of-concept for this platform and leverage the unique ability to assay genetic diversity to optimize the performance of single-input genetic switches in mammalian cells. Additionally, we show that the CLASSIC platform can be adapted to enable image-based G2P mapping of diverse features of cellular identity and phenotype, including protein compartmentalization and cell morphology, and interactions between engineered cells in multi-cellular environments, such as T cell killing
Computational Analysis and Prediction of Defect Distribution in Energy and Low-Dimensional Materials
No full textThe properties of many materials can be theoretically predicted with the assumption of an atomically perfect structure. However, this is not realistic in practice, as the presence of defects within a material’s structure can have significant effects on its properties. Visualization techniques such as microscopy and tomography enable the identification and characterization of defects within a material’s microstructure, but such analysis is often limited in scope due to the time, labor, and computation costs associated with image analysis. Computational techniques such as image processing, machine learning, and materials simulation can aid in the identification and characterization of defects, allowing for more comprehensive understanding of both the effects of defects as well as the underlying mechanisms that lead to their formation.
The first topic of this thesis presents the development of a semi-automated image processing pipeline to facilitate the construction of the largest 3D dataset of fracture within battery cathode particles, enabling statistical analysis of the factors involved in cycling-induced degradation.
The second topic of this thesis presents a novel simulation method for 2D polycrystal growth that utilizes crystal geometry to predict microstructural evolution. The simulations produced are then used in a series of statistical case studies on the effects of shape and epitaxy on crystal growth and as training data for neural networks relating early-stage crystal configuration to eventual microstructure.
The third topic of this thesis addresses limitations identified from the machine learning models and presents geometric algorithms to produce quick, reliable, and explicable predictions of current and eventual microstructures for arbitrary configurations of triangular crystals. This work supports existing research efforts to efficiently predict material quality (defect density) from images taken during the early stages of crystal growth processes for 2D material films grown via chemical vapor deposition and offers an alternative to second harmonic generation microscopy for microstructural characterization of 2D crystal films
Employing ML Methods on Digitized FOIA Requests for Improved Discoverability and Policy Research
No full textBorn-digital records pose challenges for digital preservation due to their unstructured formats and noncompliance with accessibility standards. This project introduces a modular, open-source workflow to batch process large, mixed media PDFs—many obtained through FOIA requests—by leveraging OCR, AI, and named-entity recognition. Built for the White House Scientists Archive, this system enhances discoverability and usability of digitized records across administrations and supports metadata extraction at scale. Key tools include Mistral AI for OCR, Apache Tika for entity recognition, and a finet uned Mistral model for metadata generation
Modifications and Applications of 1D and 2D Nanomaterials for the Environmental and Energy Nexus
No full textEnergy and environmental issues are among the most critical challenges in modern industrial development. For centuries, humanity has relied heavily on traditional fossil fuels such as coal, oil, and natural gas. Even today, the majority of global energy consumption is derived from these conventional energy sources, creating increasing pressure on environmental systems. Governments and industries spend substantial resources each year on environmental issues such as wastewater treatment, air quality improvement, and mitigating the effects of climate change. The pursuit of a sustainable balance between energy demands and environmental protection remains a critical challenge, driving the need for more efficient and innovative technologies in both areas.
Technology advances in areas such as electrocatalysis and water treatment are gaining traction, with a particular emphasis on the development and application of nanomaterials. In particular, 1D and 2D nanomaterials, including carbon nanotubes and transition metal dichalcogenides, exhibit unique properties such as high surface area, chemical stability, and efficient electron transport, making them highly effective in energy and environmental applications.
This thesis focuses on the design, characterization, and performance evaluation of 1D and 2D nanomaterials for sustainable energy production and environmental protection. The research includes four key projects: (1) the application of Co-doped MoS₂ for electrocatalytic ammonia synthesis and investigation of the mechanisms of catalytic performance enhancement through doping; (2) the use of Cu-doped MoS₂ for nitrate reduction to nitrogen and the analysis of atomic and elemental arrangements around the dopants; (3) the application of temperature-responsive polymers for anti-scaling treatment targeting CaCO₃ in water treatment plants; and (4) the use of pH-responsive polymers for anti-scaling treatment targeting CaSO₄. Overall, this thesis provides valuable insights into the development of advanced materials for energy and environmental applications, contributing to sustainable technological progress in these critical areas
Computational Imaging System for 3D Sensing and Reconstruction
No full textThe thesis explores three challenges in 3D imaging with different applications: 3D stereo imaging with large depth-of-field, 3D sensing with a compact device, and 3D microscopy of thick scattering samples with fast scanning speed.
The first part of this thesis focuses on a stereo imaging system that can get large depth-of-field and high-quality 3D reconstruction in light-limited environments. To overcome the fundamental trade-off between imaging volume and signal-to-noise ratio (SNR) that appears in conventional stereo, a novel end-to-end learning-based technique is proposed by introducing a phase mask at the aperture plane of the cameras in a stereo imaging system. The phase mask creates a depth-dependent yet numerically invertible point spread function, allowing us to recover sharp image texture and stereo correspondence over a significantly extended depth of field (EDOF) than conventional stereo.
The second part of the thesis exploits the strongly dispersive property of metasurfaces to propose a compact, single-shot, and passive 3D imaging camera. The proposed device consists of a metalens engineered to focus different wavelengths at different depths and two deep networks to recover depth and RGB texture information from chromatic, defocused images acquired by the system.
The third part of the thesis explores a learning-based method that can rapidly capture 3D volumetric images of thick scattering samples using a traditional wide-field microscope. The key idea is to use a 3D generative adversarial network (GAN) based neural network to learn the mapping between the blurry low-contrast image stacks obtained using a wide-field microscope and the sharp, high-contrast image stacks obtained using a confocal microscope. After training the network with widefield-confocal stack pairs, the network can reliably and accurately reconstruct 3D volumetric images that rival confocal images in terms of lateral resolution, z-sectioning , and image contrast
Entropy stable reduced order modeling of nonlinear conservation laws using discontinuous Galerkin methods
No full textReduced order models (ROMs) construct inexpensive surrogate models to reduce costs associated with many-query scenarios. Current methods for constructing entropy stable ROMs for nonlinear conservation laws utilize full order models (FOMs) based on finite volume methods (FVMs) on uniform grids. This master's thesis describes how to generalize the construction of entropy stable ROMs from finite volume FOMs to high-order discontinuous Galerkin (DG) FOMs. Significant innovations of this thesis include the introduction of new test basis involving DG weight matrix for accuracy, a dimension-by-dimension hyper-reduction strategy, and the simplification of the hyper-reduction step, which is achieved by employing the Carathéodory pruning technique specifically tailored for the hyper-reduction of boundary terms
BIOGENESIS OF NATIVE EXTRACELLULAR VESICLES AND GENERATION OF BIOENGINEERED EXTRACELLULAR VESICLES AS THERAPEUTIC AGENTS
No full textExtracellular vesicles (EVs) are small vesicles secreted from presumably all types of body cells naturally. EVs are involved in the bidirectional intercellular communication with functional impact. While the mechanism of EV generation and uptake by recipient cells is not fully understood, which is crucial for understanding their biological impact, EVs have already been considered as a drug delivery system in the context of various pathologies. To better evaluate mechanism involved in the biogenesis of EVs, I focused on the functional role of three EV-enriched tetraspanins, CD9, CD63, and CD8. Employing loss of function studies, the proteomics of cells deficient in CD9, CD63, or CD81, and EVs generated by these cells were functionally investigated. CD9, CD63, and CD81 were found to be important for sorting of specific proteins into the EVs, each one displayed distinct contribution in trafficking of proteins into EVs . Next, to explore how engineered EVs can be involved in the regulation of immunity, I designed an engineered EV-based platform for vaccine development (EVX-M+P) and for cancer immunotherapy (EVmIM). EVs were endogenously loaded with mRNA (M) and protein (P) encoding an antigen (X) for the design of EVX-M+P to induce rapid and robust adaptive immune response and protection from future exposure. As a proof of concept, spike protein of SARS-CoV-2 and human ovalbumin (OVA) were used successfully as antigens for vaccines against a viral disease and melanoma, respectively. Next, EVs were endogenously loaded to harbor multiple surface immunomodulatory proteins of CD80, 4-1BBL, CD40L, CD2, and CD32 to generate EVmIM, which could induce APCs and T cells activation simultaneously to strengthen the antigen presentation and immune response against cancer progression, validated in orthotopic melanoma mouse model. The simplicity of EVs modification and cargo loading, and successful testing of the EVX-M+P and EVmIM platforms, offer new methodologies that can streamline the development of a new class of vaccines and immunotherapies. Collectively, my thesis research opens novel vaccination and cancer immunotherapy strategies that can be developed for human testing
Interplay Between Residential Nature Exposure and Walkability and Their Association with Cardiovascular Health
No full textBackground Green space has been linked with cardiovascular (CV) health. Nature access and quality may have significant impact on CV risk factors and health. Objectives The authors aimed to investigate the relationship between NatureScore, a composite score for natural environment exposure and quality of green spaces, with CV risk factors and atherosclerotic cardiovascular diseases (ASCVD). Methods A cross-sectional study including one million adult patients from the Houston Methodist Learning Health System Outpatient Registry (2016-2022). NatureScore is a composite measure of natural environment exposure and quality (0-100) calculated for each patient based on residential address. NatureScores was divided into 4 categories: nature deficient/light (0-39), nature adequate (40-59), nature rich (60-79), and nature utopia (80-100). CV risk factors included hypertension, diabetes, dyslipidemia, and obesity. Results Among 1.07 million included patients (mean age 52 years, female 59%, Hispanic 16%, Non-Hispanic Black 14%), median NatureScore was 69.4. After adjusting for neighborhood walkability, patients living in highest NatureScore neighborhoods had lower prevalence of CV risk factors (OR: 0.91, 95% CI: 0.90-0.93) and ASCVD (OR: 0.96, 95% CI: 0.93-0.98) than those in lowest NatureScore neighborhoods. A significant interaction existed between NatureScore and Walkability (P < 0.001), where those in high NatureScore (≥60) high walkability (≥40) areas had lower prevalence of CV risk factors (OR: 0.93, 95% CI: 0.90-0.97, P < 0.001) and were more likely to have optimal CV risk profile (relative risk ratio: 1.09, 95% CI: 1.04-1.14, P = 0.001). Conclusions These findings suggest that while green spaces benefit health, their accessibility through walkable environments is crucial for cardiovascular disease protection
Use of Predatory Lending Products in the Houston Area
No full textThe Houston area is an interesting location for studying the use of predatory lending products because it is both economically prosperous and has many residents living at or below the poverty line. This study uses survey data from the Greater Houston Community Panel (GHCP) to examine the prevalence of predatory lending product use and the economic and financial factors related to residents' use of them, such as a person's access to credit and experiences of financial stress. It aims to inform targeted solutions that promote economic inclusion, reduce reliance on high-risk financial products, and build long-term financial security for more residents in the Houston area. Key findings: About 1 in 5 Houston-area residents used a predatory lending product in the past year; over half of residents who used a predatory lending product said they needed the money to cover basic necessities such as food, groceries, or housing expenses; residents who had been turned down for a loan had a 29% chance of using predatory lending products, compared to a 17% chance for those who were approved for the full amount; residents with lower credit scores had higher probabilities of using predatory lending products than those with higher credit scores; greater difficulty paying for housing is related to a higher likelihood of using predatory lending products; and residents with at least 3 months' worth of savings to cover living expenses, a key marker of financial security, were significantly less likely to use predatory lending products than those with less savings
Acoustically-Targeted Delivery of Engineered Vectors to the Brain
No full textThe brain is connected through neuronal pathways that together turn individual neurons into the brain. Targeted gene delivery to the brain is a critical tool for neuroscience research and has significant potential to treat human disease. However, the site-specific delivery of common gene vectors such as adeno-associated viruses (AAVs) is typically performed via invasive injections, which limit its applicable scope of research and clinical applications. Alternatively, focused ultrasound blood-brain-barrier opening (FUS-BBBO), performed noninvasively, enables the site-specific entry of AAVs into the brain from systemic circulation. However, when used in conjunction with natural AAV serotypes, this approach has limited transduction efficiency and results in substantial undesirable transduction of peripheral organs. Recent discovery of viral vectors that can selectively transduce neuronal projections have enabled discovery of how various brain regions communicate to affect behavior, but they also suffer from delivery challenges. Here, we evaluate the potential of two engineered viral vectors, AAV.FUS and AAV9.retro, for noninvasive site-specific transduction of the neurons and neuronal projections after systemic delivery. The resulting vectors substantially enhance ultrasound-targeted gene delivery, neuronal tropism and transduction of neuronal pathways while reducing peripheral transduction. In addition to enhancing the only known approach to noninvasively target gene delivery to specific brain regions, these results establish the ability of AAV vectors to be evolved for specific delivery applications and therapies