Abstract Withheld.

Speckle is an imaging artefact that arises from interference in high-coherence sources, such as conventional single-mode lasers. Thus, one approach to removing speckle in imaging is using spatially incoherent multi-mode lasers. For this reason, we are interested in achieving multi-mode lasing in the terahertz frequency range – a frequency range that is promising for imaging applications due to its non-ionizing yet penetrative nature. Using quantum-cascade vertical-external-cavity surface-emitting lasers (QC-VECSELs) – lasers in which in one reflecting end of the laser cavity is combined with metal-topped ridges that are filled with quantum-well gain material – our lab has previously achieved single-mode terahertz lasing. However, multi-moding is more difficult to achieve due to gain competition and spatial hole burning. In this work, we present a 2D model of QC-VECSEL metasurfaces with random ridge widths, based on our hypothesis that randomizing the ridge widths will localize lasing modes to different areas of the metasurface, reducing spatial hole burning. To investigate the validity of this approach, we generated various random metasurfaces and simulated their cavity fields using COMSOL Multiphysics. We then extracted all modes lasing under a fixed gain threshold and calculated their mutual spatial overlap across all biased ridges. In these simulations, we demonstrated that randomizing the metasurface increased the number of lasing modes and decreased spatial overlap significantly, at the cost of increased thresholds. In addition, uncertainty remains in how accurately our 2D models will represent 3D devices, and in how much we actually need to reduce spatial overlap for successful multi-moding.

Recently, soft materials have been under increasing study due to their high flexibility. We are studying a soft material known as liquid crystal elastomers (LCEs). With the ability to change shape when exposed to heat or light, LCEs have found applications in artificial muscles, biomedical devices, and more. These materials are made of liquid crystals (LCs) embedded in the backbone of a polymer, and their LCs can reorient in the direction of an applied stress. Our goal is to experimentally study the mechanical properties of LCEs and further explain their constitutive behavior by focusing on LCEs whose LCs are oriented in the same direction, called monodomain LCEs. We conducted relaxation tests and uniaxial stress-strain measurements at different loading rates on monodomain LCEs that we fabricated with different LC orientations. Effects of stress on LC reorientation were measured using transmission circular polariscopy (TCP), which characterizes the rotation of LCs during deformation. In addition, digital image correlation (DIC) was used to study the shear strain distribution in stretched LCEs. Our results show that the stress-strain relation of LCEs highly depends on loading rate and LC orientation. At low strain rates, the hysteresis in stress-strain curves is small, and LCs can reorient with strain, while at high rates, the hysteresis is large, and there is a lag of LC reorientation. Also, our DIC tests show that LC reorientation causes shear strain within LCEs. These experiments further contribute to the development of theoretical models of LCEs and drive forward LCE technologies.

Targeted therapies eliminate cancer cells by inhibiting specific dysregulated pathways. While these treatments have proven to extend and save lives, drug resistance is a prevalent issue for currently available drugs. Previous work has identified sets of genes in melanoma that are disproportionately expressed in cancer cells that go on to become resistant after treatment with targeted therapy. However, how these genes form a coordinated pathway is not understood. We used a previously proposed non-linear form of gene regulatory network identification to convert perturbation experiments into an inferred pathway for these genes. Extending this work, we developed an iterative matrix solving method, making this algorithm scalable to many thousands of genes and knockdown conditions, along with allowing us to reason about its statistical properties. We applied this model and then analyzed its results to reveal drivers of melanoma drug resistance development. The matrix was visually represented by a weighted directed diagram, which was analyzed for clustering using the Bellman-Ford distance algorithm for pairs of pre-resistant, resistant, and randomized nodes. In total, this work provides a scalable approach to reasoning about the pathway mechanisms revealed in perturbation experiments.

Liquefaction occurs when a high water content soil loses strength during an earthquake and enters a liquid-like state. Sea level rise will elevate local beach groundwater tables and may substantially increase liquefaction risk along the California coastline. Studies have been conducted in Hawaii and Connecticut to understand the combined risk of liquefaction and sea level rise, however studies for the California coastline are limited, as are the publicly available groundwater well data required to make these risk assessments. This research highlights liquefaction risk within specific sites, how sea level rise affects these risks, and considers the impact at various urbanized coastal California sites. Sunset Beach and Newport Beach serve as case studies. Information such as population, number of buildings, and building type are compiled. Geospatial surface terrain data and tidal elevations are then compiled in ArcGIS to create elevation maps for the sites and used to make a general assessment of potential liquefaction damage. For the Sunset Beach site the areas currently within liquefaction zones include 403 buildings and $849 million in housing while Newport Bay includes 6,934 buildings and $26.6 billion in value. Limited groundwater data suggests that groundwater tables within the sites are at a moderate depth. This, combined with the extent that SLR could temporarily increase groundwater levels at the sites through direct inundation suggests that these coastal communities may become increasingly susceptible to liquefaction. Study results emphasize the severity of liquefaction risks and highlight the need to monitor local beach groundwater tables and potential liquefaction conditions to reduce current and future threats to infrastructure.

Precision farming is a hotly pursued field in modern agriculture that seeks to target individual weeds as opposed to entire fields, thereby dramatically reducing the application of herbicide and mitigating the negative impact of herbicides on human and environmental health. In this project we present an autonomous weed-spraying robot which utilizes computer-vision-based navigation and weed-identification algorithms in order to spray weeds with high efficiency for multiple stages of plant growth. This compact platform is the first to autonomously spray row crops such as flax and canola which have row spacings under 30cm, and offers an inexpensive alternative to larger industrial robots designed for large row spacings. The robot identifies flax plants using an implementation of the YOLOv5 neural network, and uses the plant locations both to navigate and to spray weeds. Furthermore, a novel self-charging station design and sprayer system allows the robot to operate for the entire day. This robot was tested in the flax fields surrounding the North Dakota State University. Preliminary results indicate that the weeds are successfully killed in the center of rows, but they persist immediately next to the crop-line where the robot cannot access without trampling the flax. Additionally, further development of the autonomous navigation and weed identification algorithms is needed for dependable performance in the field. However, overall our work provides a strong basis on which to build a robust platform that could substantially decrease the amount of herbicide required to grow crops, and increase the health of both humans and the environment.

Duchenne Muscular Dystrophy (DMD) is a rare genetic neuromuscular disease that affects 1 in 3500 male live births. Caused by mutations in the DMD gene, DMD is characterized by progressive muscle weakness due to deficiencies in dystrophin, a sarcolemma protein that maintains muscle integrity during contraction. Although DMD currently has no cure, previous studies have proposed viral vectors that can deliver gene-editing machinery to dystrophin-deficient muscle cells. However, a more permanent treatment is needed to correct mutated DMD genes since these recombinant viral vectors dilute after cell division. To improve the efficiency of adeno-associated virus 9 (AAV9) vectors and facilitate cell type-directed gene transfer, we generated a randomized peptide display library of AAV9 variants for myocytes and muscle stem cells (MuSCs). By screening these tissue-enriched variants and characterizing them with Sanger sequencing, our library identifies AAV9 surface proteins that demonstrate skeletal muscle and muscle stem cell transduction. With structural and functional analysis of sequenced surface proteins, we conclude that our AAV9 library can be a useful tool for optimizing cell-specific transduction and selecting skeletal muscle- or MuSC-targeting vectors for future DMD gene therapies.

Chronic inflammatory demyelinating polyneuropathy (CIDP) is a debilitating autoimmune disease of the peripheral nervous system that can have favorable outcomes with prompt treatment. Despite these therapies, diagnosis of CIDP remains challenging and nonspecific, subjecting patients to extensive and potentially inconclusive tests such as nerve conduction studies and CSF examination. The presence of autoantibodies has been used for diagnosis of autoimmune diseases such as type I diabetes and systemic lupus erythematosus. The antibody profiles of CIDP and healthy control patients have been characterized by collaborators at UCSF using phage immunoprecipitation sequencing (PhIP-seq). Our analysis of these data suggest several autoantibodies as potential diagnostic markers for CIDP. Since the PhIP-seq discovery platform lacks specificity we are using the luciferase immunoprecipitation system (LIPS) to validate the presence of autoantibodies in CIDP. Using the LIPS, an antigen is fused to Renilla luciferase (Ruc) which produces luminescence in the presence of coelenterazine. We have fused interferon alpha (IFNɑ) to Ruc and have collected sera from patients with autoimmune polyglandular syndrome type 1 (APS-1) which is known to contain anti-IFNɑ autoantibodies. Upon addition of the fusion IFNɑ-Ruc protein to APS-1 patient sera we expect anti-IFNɑ autoantibodies will bind the fusion protein. After precipitation of antibodies from the patient sera, coelenterazine is added and the sample luminescence is measured. Since the concentration of luminescence is dependent on the concentration of IFNɑ-Ruc protein, luminescence is a proxy for the concentration of anti-IFNɑ autoantibodies. Ultimately we hope to find autoantibodies specific to CIDP patients and improve our diagnostic capabilities.

Wetland environments are home to biological processes that emit methane - a greenhouse gas that accelerates climate change. Methanogens, methane-producing bacteria, are commonly present in these wetlands that receive stormwater runoff containing a range of contaminants including microplastics (MPs). MPs are known to suppress microbial activity, but their effect on the methanogenesis process and on the diversity of microbial communities in wetland environments remains unknown. To identify the impact of MPs on the methane production from wetland sediments, we performed a series of laboratorial batch experiments where different wetland sediment sizes were exposed to MPs under aerobic and anaerobic conditions while monitoring the production of methane. The batch experiments were kept in incubators at optimum conditions for methane production. Our results show that, under aerobic conditions, large sediments (> 2.0 mm) produced 100 times more methane than fine sediments (< 2.0 mm), possibly due to the decomposition of organic matter present only within large sediments. Under anaerobic conditions, fine particles increased their production by more than 40 times when acetate was present in solution, proving that wetland sediments have the capacity for methane production when nutrients are available. The presence of MPs among fine sediments with acetate caused a delay and suppressed methane production, possibly due to the direct interaction between MPs and methanogens, as well as the indirect interaction between MPs and acetate, as both processes could suppress methane production. Overall, the results furnish the understanding of methane fluxes in relation to microplastic transport in wetland environments.

Multiple studies highlight differences in daily sleep and wake patterns between men and women yet there is minimal evidence outlining how biological sex directly affects sleep regulatory processes. Previous studies determined that sex differences in recovery sleep amount are partially dependent on sex chromosomes. To examine spontaneous sleep states and the ability to recover from sleep loss, polysomnographic sleep recordings were obtained from gonadectomized (GDX) FCG mice. Recovery from sleep loss demonstrated sex differences in sleep amount partially driven by sex chromosome complement. The XY* mouse model enables investigation of these sex chromosome regulation by varying genetic dosages of the X or Y sex chromosomes. The XY* line is comprised of four sex chromosome complement phenotypes: phenotypic female (XO/XX), phenotypic male (XXY/XY*). Adult mice were implanted with electroencephalograph (EEG) and electromyography (EMG) recording electrodes. Mice were subjected to 24 hrs of baseline (spontaneous) sleep-wake recording, after which they’re subjected to six hours of sleep deprivation by gentle handling, followed by recovery until completion of the 24-hr recording period. Both data in Recovery Active phase total sleep and NREM show statistical significance when the Y chromosome is absent. Immediate early gene signaling was then performed in sleep-active neurons in the hypothalamus through immunohistochemistry as supplemental data to these results. Identifying Fos cells serves as a neuronal activation marker after sleep deprivation has been administered for each genotype in the XY* model. These results postulate that genetic factors on the sex chromosomes encode varying homeostatic responses to sleep loss.