48670 research outputs found
Sort by
Development of Novel Vaccination Strategies Against Emerging Bunyaviruses
No full textSevere fever with thrombocytopenia syndrome virus (SFTSV) is a highly pathogenic emergent bunyavirus. First isolated in 2009 in China, SFTSV is now endemic to several east Asian countries where high case fatality ratios of 6-30% are reported. The primary tick vector of SFTSV, Haemaphysalis longicornis, has a large range and is a well reported invasive species throughout the world. This, in addition to SFTSV’s ability to spread in the absence of its vector in nosocomial and veterinary settings, suggests SFTSV is well suited to cause widespread lethal outbreaks. Currently no vaccines or therapeutics against SFTSV exist, prompting health agencies to list SFTSV as a high priority pathogen. Here, we first develop a single dose recombinant vesicular stomatitis virus (rVSV) encoding the SFTSV glycoproteins Gn/Gc as a virus vectored vaccine. We demonstrate that this vaccine (rVSV-SFTSV) is safe in immunocompromised mice and not neuropathogenic when delivered intracerebrally. Additionally, this vaccine induces robust antibody responses that are protective from lethal challenge. Furthermore, we demonstrate that this vaccine elicits cross-protective responses against the closely related Heartland virus. We then developed an mRNA vaccine encoding SFTSV Gn/Gc and compared this platform with our rVSV-SFTSV vaccine in single dose, homologous prime-boost, and heterologous prime-boost regimens. We found that mRNA immunizations in single dose and homologous prime-boost regimens achieved the highest neutralizing antibody titers. Immunizations with rVSV-SFTSV also reached high antibody titers though they were the lowest titers of any immunization regimen, with heterologous prime-boost having intermediate titers. When T-cell responses were analyzed, mRNA immunization achieved robust CD4+ and CD8+ responses in single dose and homologous prime-boost regimens. Heterologous vaccine regimens elicited similar responses to homologous mRNA strategies despite weak cellular activity after rVSV-SFTSV prime. Despite some differences in immunogenicity, all vaccines were protective from lethal SFTSV challenge. Overall, this work demonstrates the effectiveness of two vaccine platforms in their ability to elicit robust protective responses against SFTSV
Strategies of Technology Adoption for Managing IT Skilled Labor
No full textEmployers repeatedly reevaluate their human capital investments when new technologies enter the market. Adopting new technologies can afford firms to achieve productivity gains (Schumpter 1942, Abernathy and Clark 1985, Anderson and Tushman 1990) while offering workers the opportunity to update their skillsets on the job (Tambe et. al 2020). However, there are risks to both the employer and the worker as managers often struggle to invest in the appropriate innovations or match pace with the response to new technologies from workers (Cooper and Schendel 1976, Henderson 1993, Benner 2010). This dissertation addresses how employers can best adapt their technology strategies given the preferences of IT skilled workers. In the first chapter, we offer insight into the considerable heterogeneity in demand for technical skills across different employers, even within the same job role. We develop a model of investment in technology mix as a source of differentiation among employers\u27 specific human capital where heterogeneity in the choice of technologies can arise in equilibrium based on heterogeneous employer characteristics. Using data on over 5 years of job vacancies, we characterize the technology choices made by over 1,700 employers. We find that large employers have greater investment into firm-specific technologies while also deriving more payoff from pursuing standard technologies. Smaller firms benefit more from firm-specific technologies but are constrained by more limited ability to invest. These results are unique to IT skills, and we do not find the same to hold for all skills in general. In the second chapter, we explore a case study of emerging technologies in the context of open-source software (OSS) sponsored by corporate entities. Corporate OSS informs on how employers manage potential external communities of talent. We create a novel dataset of open-source contribution records to study the effect of star developer contributions on corporate OSS projects and find that contributions by star developers result in about 52% more in contributions and 24% less in the hazard of becoming inactive on the GitHub social coding platform. Furthermore, projects benefit indirectly by incorporating external dependency packages with star contributions and are most successful when star contributions are paired with a team of supporters. These results suggest that open-source is a community inclined to a few highly influential actors rather than a crowd of anonymous contributors. Quantifying the influence of these stars have implications for how a platform can stimulate participation in certain areas of interest and which strategies companies can deploy for successfully engaging with the open-source community. In the third chapter, we derive a matching model between workers and employers based on their heterogeneous investments into a technology. Our model predicts the commonly observed positive assortative matching phenomenon wherein higher ability workers are matched to higher ability employers in equilibrium. Our model predicts that higher ability workers and firms are most susceptible to breaks in matching under a shifting technology frontier. We additionally find empirical evidence that higher ability firms more quickly change their skill requirements. These results have implications for understanding inequality in reskilling and retention of the workforce as they depend on the discrepancy in abilities between worker and employer
Knowledge-Guided Deep Learning Models of Drug Toxicity Improve Interpretation
No full textIn drug development, a major reason for attrition is the lack of understanding of cellular mechanisms governing drug toxicity. Conventional models are limited by low accuracy and lack of interpretability. Further, they often fail to explain cellular mechanisms underlying structure-toxicity associations. To address these limitations, we developed a series of interpretable in silico models that connect drugs to their toxicity targets and pathways. In Chapter 2, we incorporated target profile as an intermediate connecting structure to toxicity. To accommodate for high-dimensional feature space, we developed a pipeline named TargetTox that can identity subset of predictive features. The features identified by TargetTox accurately predicted binding outcomes for 377 targets and toxicity outcomes for 36 adverse events. We demonstrated that predictive targets tend to be differentially expressed in the tissue of toxicity. We also discovered that predictive targets are enriched for key cellular functions associated with toxicity. Furthermore, we found evidence supporting diagnostic/therapeutic applications of some predictive targets. Our findings highlighted the critical role of predictive targets in cellular mechanisms leading to toxicity. In Chapter 3, we developed DTox, an interpretation framework for knowledge-guided neural networks, which can predict compound response to toxicity assays and infer toxicity pathways of individual compounds. We demonstrate that DTox can achieve the same level of predictive performance as conventional models with a significant improvement in interpretability. Using DTox, we were able to rediscover mechanisms of transcription activation by nuclear receptors, recapitulate cellular activities induced by aromatase inhibitors and PXR agonists, and differentiate distinctive mechanisms leading to HepG2 cytotoxicity. Virtual screening by DTox revealed that compounds with predicted cytotoxicity are at higher risk for clinical hepatic phenotypes. In Chapter 4, we introduce AIDTox, an interpretable deep learning model which incorporates curated knowledge of chemical-gene connections, gene-pathway annotations, and pathway hierarchy. AIDTox can accurately predict cytotoxicity outcomes. It also provides comprehensive explanations of cytotoxicity covering multiple aspects of drug activity including target interaction, metabolism, and elimination. In summary, our work provides a framework for deciphering cellular mechanisms of toxicity in silico
Modeling Infertility: Single Gene Mutations and Consequences for Germ Cell Development
No full textInfertility is a growing global healthcare concern. Genetic causes account for a large percent of infertility cases, and in some instances, single-gene mutations are sufficient to cause infertility in humans. To further our understanding of certain single gene causes of infertility, we studied the function of three conserved genes, Trip13, Cyclin B3, and Flacc1, using knockout mouse lines. Trip13 is an ATPase that remodels HORMA-domain proteins and is essential for meiotic progression. Pathogenic mutations of TRIP13 are found in humans and cause infertility via oocyte maturation arrest. In Chapter 2, we used a genetic approach and imaging techniques to study the function of TRIP13 in meiotic prophase I in mice. We found that TRIP13 localizes to the synaptonemal complex (SC) in pachynema and to meiotic telomeres throughout prophase I. Trip13-/- spermatocytes arrest in pachynema and exhibit defective synapsis. TRIP13 localization in other mutant strains with defective synapsis, indicated that TRIP13 is likely recruited by a component of the SC during early synapsis. In Chapter 3, we examined CCNB3, a meiosis-specific cyclin, which is required for meiosis I in mice and for zygotic genome activation (ZGA) in Ciona intestinalis. In humans, CCNB3 mutations cause recurrent pregnancy loss. We explored meiotic and post-meiotic roles of CCNB3 using imaging studies, fertility tests, and pronuclei swap experiments. Ccnb3-/- females are sterile. Ccnb3-/- oocytes cannot complete meiosis I but can be fertilized. Resultant embryos are triploid, resulting in embryonic lethality. Cytoplasmic Ccnb3 was not required for embryonic development, and its function in ZGA is not conserved. In Chapter 4, we used varied approaches (microscopy, mating tests, histology, Western blot), to explore the function of the coiled-coil protein FLACC1. FLACC1, a testis-specific protein, was thought to be an important component of the sperm tail. However, Flacc1-/- males were fertile. FLACC1 Western blot showed testis specific expression, yet previous reports of localization to the sperm tail were not confirmed. Together, this work provides insights into the germ cell-specific function of three conserved genes, two of which are required for mammalian fertility
Robustness of Temporal Logics with Applications to Safe Autonomy
No full textSignal Temporal Logic (STL) is a common way to express a broad range of real-time constraints that can be imposed on control systems. Spatial robustness of STL specifications, quantifying permissible spatial perturbations, has been widely studied in the literature. However, despite the importance of various time-critical systems, temporal robustness of STL has not yet been studied in depth nor has been used for control design. In the first part of this thesis, we establish a comprehensive theoretical framework for temporal robustness of STL. We show that temporal robustness quantifies the extent to which timing uncertainties can be tolerated without violating real-time specifications. We define synchronous and asynchronous temporal robustness and show that these notions quantify the robustness with respect to synchronous and asynchronous time shifts in the predicates of the underlying signal temporal logic specification. We further prove that synchronous temporal robustness upper bounds asynchronous temporal robustness. Moreover, we show under which conditions these two robustness notions are equivalent. Introduced synchronous and asynchronous notions are directional and consider either left or right perturbations. Due to this reason we additionally define and study the combined temporal robustness which simultaneously considers left and right time shifts. In the second part of this thesis, we focus on applications of various robustness functions derived in the first part of the thesis to robust planning and control design questions. We first propose solutions to temporally-robust control synthesis problem by presenting Mixed-Integer Linear Programming (MILP) encodings for derived temporal robustness notions. Second, we solve the spatially-robust control synthesis problem and show how to adapt the smooth operator for space robustness maximization. Furthermore, we propose possible distributed solutions to centralized multi-agent planning problems. Through various simulations, as well as experiments on actual robotic systems, we showthat our presented solutions are computationally efficient as well as can be used in a wide variety of applications
Resilient Information Theoretic Active Exploration for Multi-Robot Teams
No full textOver the past decades we have seen robots move from constrained and heavily designed industrial environments out into the unstructured world. This shift drives a need for smaller, safer, and less expensive robots which can collaboratively complete tasks autonomously. For such teams to function, they must be able to reach a shared understanding of their task and environment while accommodating unreliable sensors, and also to recover gracefully from individual failures without requiring centralized coordination. These challenges are the focus of this thesis which lays the groundwork for multi-robot active mapping that is resilient to faulty or malfunctioning sensors. Such algorithms have wide ranging applicability, from persistent monitoring tasks such as those found in agriculture to time critical tasks such as search and rescue. First we present two map representations designed specifically for autonomous information theoretic mapping. For each method, we develop an information theoretic value function which can be used to choose actions to maximize the information gained about the map. We present a principled method for accounting for both information gained by exploring new areas, as well as information gained by further inspection of the existing map to account for sensor uncertainty. To address the vulnerability of local planning methods to local minima, we develop a strategy to maintain a long horizon planning tree over time. Second, we extend a highly distributed approach to resilient consensus for static networks to applications with multi-robot teams. This approach has been largely limited to small static networks or strict formations. First we develop a method which can be used for teams with time-varying range-based communication which is suitable for tasks where robots are not required to spread out in the environment. We then present a method that is well suited to mapping and coverage applications which uses a well known communication structure to guarantee successful resilient consensus. Finally we present examples of how these tools can be used to enable resilient active mapping and coverage for teams of robots with faulty or malfunctioning sensors
Multibreath Hyperpolarized Gas Imaging of Lung Ventilation and Gas Exchange in Humans for Diagnosis and Treatment Response Monitoring
No full textMagnetic Resonance Imaging using hyperpolarized helium and xenon gas (HP gas MRI) is uniquely sensitive to aspects of lung function integral to that organ’s ability to sustain life. The mixing of freshly hyperpolarized inspired gas with gas already present in the lung can be directly assessed using HP gas MRI—by visualizing not just the inhalation of the contrast agent, but also its diffusion throughout the acinar structure and, in the case of xenon, its dissolution into tissue, binding to hemoglobin, and distribution to other organs via the circulatory system. In this way, the dynamics of HP gas in the lungs mirrors that of oxygen and carbon dioxide. Identifying barriers to gas exchange is central to understanding which aspects of a diseased lung are most tied to morbidity, and whether available treatments can resolve those deficits. While regional lung function abnormalities can be masked by clinical global metrics, HP gas imaging is capable of addressing the spatial scale of these inhomogeneities for regional minute ventilation, residual volume, local exchange with lung parenchyma tissues, and with that of red blood cells, all at a satisfactory resolution. This dissertation presents ‘multibreath’ approaches that involve imaging over a series of breaths—in which a single, rapid image is acquired during a short breath-hold at the same phase of the breathing cycle on each breath (end-inhale or end-exhale). Different image contrasts may be encoded to highlight various features (e.g., diffusion, gas exchange, etc.), but the overall goal is to assemble all of the images to better capture the complex dynamics of gas delivery in the functionally compromised lung during the relatively normalized—i.e., tidal—breathing that better reflects the limitations encountered in subjects’ daily life. This approach can distinguish abnormally slow-filling regions that may require several breaths to resolve, perhaps through partially blocked airways or collateral pathways, from those that are entirely nonfunctional. This distinction is important, as slow-filling parenchyma can comprise more than half of the lung volume in cases of severe disease, and must therefore be responsible for significant life-preserving gas exchange. Our multibreath approach for imaging lung ventilation and gas exchange is applied here to the diagnosis, classification, and monitoring of treatment response in chronic obstructive pulmonary disease (COPD). COPD encompasses several pulmonary disorders, including chronic bronchitis and emphysema, and is characterized by both altered lung structure and function. The obstruction associated with COPD is likely, at least to some extent, a disruption of balanced forces maintaining airway patency in the healthy lung. However, the root causes of structural and functional changes may differ among the various presentations of the disease. Because COPD is a complex disorder with multifaceted manifestations, it is an ideal disease model on which to test the efficacy of the functional imaging tools developed in this dissertation for both early diagnosis as well as monitoring both disease progression and treatment response. We propose that, in both diagnosing and assessing the response to an intervention, probing the deviation of functional inhomogeneities from that of healthy nonuniformity can reduce clinical uncertainty for each patient as well as elucidate treatment changes such that these effects can be emphasized in future therapeutic advances. We further believe that observing longer term alterations leading to loss of functional benefit can help to select patients and procedures yielding persistent quality of life gains. After establishing the multibreath imaging methodology and testing its short- and long-term reproducibility, the sensitivity of the imaging markers for detecting difficult-to-assess smoking-induced subclinical changes in asymptomatic smokers is evaluated and compared with that of current clinical pulmonary assessments. Next, the effectiveness of imaging-based markers in predicting future lung function decline in smokers was compared to current clinical diagnostic techniques. We then imaged COPD patients who underwent an expensive regional treatment, endobronchial valve placement, with the aim of localizing and quantifying the most probable contributors to both functional and subjective benefit. Finally, the last chapter presents a novel ‘dynamic’ imaging approach in which subjects are imaged continuously during five minutes of spontaneous tidal breathing to generate maps of lung function capable of elucidating transient gas flow defects, pulmonary vascular response to the heartbeat, and the lung’s contribution to systemic bloodflow—aspects of lung physiology and pathophysiology that are otherwise invisible
Maladaptive Patterns of Stress Responding in Vulnerable Populations
No full textHow people respond to stress influences their risk for psychological maladjustment, especially anxiety and depression. Identifying patterns of stress responding that may be problematic, as well as vulnerability factors for engaging in these response styles, is important for informing appropriate prevention and treatment. Chapter 1 investigated the relationship between stress sensitivity, a transdiagnostic risk factor for anxiety and depression, and difficulties disengaging from stressful stimuli. We enrolled a college student sample given rising concerns over stress-related mental health problems in this population. We created and validated a 400-word novel lexical stimulus set specific to college students. Using this stimulus set in a modified dot-probe task, we showed that stress sensitivity is associated with quicker disengagement from all stimuli, not just stress stimuli, in stressful contexts. This work highlights the complex nature of attentional bias toward stress and provides a novel tool for further investigation of stress responding in the increasingly vulnerable college population. Chapter 2 used ecological momentary assessment to investigate inert (persistent) responding to stressful events in daily life. In a mixed sample of adults with generalized anxiety disorder and major depressive disorder, we examined how long the affective and cognitive responses experienced immediately after a stressful event predicted subsequent responding. Stress response inertia was evident for a period lasting up to 4.5 hours after a stressful event. Individuals with higher trait negative emotionality and greater anxiety and depression severity exhibited more prolonged stress responding. Collectively, these studies demonstrate the importance of examining stress responding beyond reactivity alone, identify speed of disengagement and response inertia as important patterns of stress responding associated with markers for maladaptive anxiety and depression, and offer clinical considerations in light of these findings
Leveraging System Call Interposition for Low-Level Process Manipulation
No full textModern software continues to grow in size and complexity with no signs of slowing down. Program tracing allows us to observe the execution of a program. OS-level program tracing is useful, as it allows us to abstract over many details of program execution and view programs based on the IO operations they perform. Beyond read-only program tracing, this dissertation overviews low-level process manipulation. We argue process manipulation is a useful and general technique with many applications. We show the utility of tracing and process manipulation by covering several projects which leverage these techniques. First, we describe DetTrace, a deterministic container abstraction. DetTrace provides a containerized environment where any computation inside the container is guaranteed to be deterministic. Next, we describe ProcessCache implements a system for automatically caching results of process-level computations. ProcessCache automatically infers inputs and outputs to a program and will only re-execute a process if its inputs have changed. Otherwise, it skips unnecessary recomputation by using the cached results. Finally, Tivo combines lightweight determinism enforcement with record and replay to suppress certain types of thread-level nondeterminism. Finally, our future work proposes ChaOS, a fuzzing system for fault injection at system call sites. Lastly, we list key features and requirements for next generation program tracing and low-level process manipulation
Dolutegravir in Neurons and Macrophages: Molecular Mechanisms of Neurotoxicity and Neuroprotection
No full textDespite effective antiretroviral (ARV) therapy (ART), human immunodeficiency virus (HIV)-associated neurocognitive disorders (HAND) persevere, affecting up to 30%-50% of all HIV-infected individuals. Although the end result involves neuronal damage and death in the absence of ART, neurons themselves are not directly infected by HIV; therefore, the pathological mechanisms are most likely indirect. Biomarkers of inflammation, immune activation, and oxidative stress remain elevated in the cerebrospinal fluid (CSF) of HIV-infected individuals throughout the course of infection regardless of the ART status. HAND is multifactorial, and both the presence of infected and activated macrophages, which are considered as a major source of HIV replication in the central nervous system (CNS), and ARV-mediated toxicity have been implicated in the persistence of HAND. Dolutegravir (DTG)-containing regimens are first-line treatment options recommended by many health guidelines. However, several studies have raised concerns regarding DTG-associated CNS abnormalities. Taken together, we hypothesize that DTG contributes to neuronal death directly via oxidative stress induction, while indirectly providing neuroprotection from HIV-induced neurotoxicity by inhibiting viral replication in macrophages. In this dissertation, we have identified two related yet independent mechanisms of DTG-mediated neurotoxicity, each of which may contribute to HAND pathogenesis. First, we show that DTG mediates neurotoxicity in vitro via increased reactive oxygen species (ROS) production and oxidative stress induction, as indicated by the alleviation of DTG-mediated neurotoxicity with the inducers of endogenous antioxidant response. Second, we generated supernatants from monocyte derived macrophages (MDM) exposed to DTG, with and without HIV infection or HIV/MDM and Mock/MDM, respectively. And we demonstrated that DTG-treated HIV/MDM has lower glutamate released compared to HIV-only/MDM. Moreover, DTG-treated HIV/MDM supernatants were not neurotoxic, compared to HIV/MDM supernatants. On the other hand, in the absence of HIV, DTG-treated Mock/MDM have increased expression of nuclear factor (erythroid-derived 2)-like 2 (NRF2), a transcription factor that regulate the expression of antioxidant and cytoprotective genes. To our knowledge, this is the first study looking at both neurons and macrophages simultaneously. Taken together, DTG may contribute to the pathology seen in HAND patients both directly and indirectly. From a translational point of view, these findings not only impact ARVs prescription based on possible adverse effects, but also highlights the importance of both targeted- drug therapy and adjunctive therapies