UBIR Repository (Univ. at Buffalo)
Not a member yet
5736 research outputs found
Sort by
Regulation of the Hippo Signaling Pathway in Mammary Epithelial Cells
No full textPh.D.The Hippo signaling pathway is an evolutionarily conserved pathway originally discovered in Drosophila melanogaster and later found to have human orthologues. Regulation of this pathway is tightly controlled throughout development and in maintaining tissue homeostasis. Upstream regulators and external signaling nodes that converge upon this pathway to control Hippo gene activity has been extensively studied in both normal development and disease. Upon external stimuli from GPCRs, cell adhesion proteins, and various other signaling pathways a kinase cascade occurs. MST1/2 are phosphorylated and activated, resulting in the interaction with scaffold protein Sav1. This complex then induces the phosphorylation of LATS1/2, leading to activation and interaction with PTPN14 and KIBRA. LATS1/2 phosphorylates the key effector proteins of the pathway, YAP/TAZ, leading to their cytoplasmic retention or proteasomal degradation. Dysregulation of upstream regulators leaves YAP/TAZ in an un-phosphorylated state and allows their translocation to the nucleus. YAP/TAZ act as transcriptional co-activators of the TEAD family of transcription factors. The YAP-TAZ-TEAD complex drives transcription of genes associated with cell proliferation, cell cycle progression, epithelial to mesenchymal transition, metastasis, and stemness. While this pathway has been well established in cancer there are still many questions that need to be answered. KIBRA is a member of the WW domain-containing protein family and has recently been reported to be an upstream protein in the Hippo signaling pathway. The clinical significance of KIBRA deregulation and the underlying mechanisms by which KIBRA regulates breast cancer (BC) initiation and progression remain poorly understood. Here, we report that KIBRA knockdown in mammary epithelial cells induced epithelial-to-mesenchymal transition (EMT) and increased cell migration and tumorigenic potential. Mechanistically, we observed that inhibiting KIBRA induced growth factor-independent cell proliferation in 2D and 3D culture due to the secretion of amphiregulin (AREG), an epidermal growth factor receptor (EGFR) ligand. Also, we show that AREG activation in KIBRA-knockdown cells is dependent on the transcriptional coactivator YAP1. Significantly, decreased expression of KIBRA is correlated with recurrence and reduced BC patient survival. In summary, this study elucidates the molecular events that underpin the role of KIBRA in BC. As a result, our work provides biological insight into the role of KIBRA as a critical regulator of YAP1-mediated oncogenic growth and may have clinical potential for facilitating patient stratification and identifying novel therapeutic approaches for BC patients. USP1 is a deubiquitinating enzyme that allows for the removal of ubiquitin from target substrates, leading to increased protein stability or altered cellular trafficking. The role of USP1 has been well established in DNA damage response in multiple pathways including the Fanconi Anemia pathway, trans-lesion synthesis, and homologous recombination, however, how it regulates other genes is still under investigation. Herein we describe how loss of USP1 can alter TAZ protein stability leading to reduced proliferation. Mechanistically, we show that loss of USP1 reduces TAZ protein levels in both non-transformed and transformed mammary epithelial cells, without altering TAZ mRNA levels. Inhibition of the proteasome can rescue TAZ loss in the absence of USP1. Also, treatment of stable USP1 knockdown cells with cycloheximide showed a reduction in the half-life of TAZ proteins. Using ubiquitin-based immunoprecipitation assays we show that loss of USP1 can increase TAZ ubiquitination and this ubiquitin modification may occur at lysine 45 and 46 within TAZ. We further mapped the interaction domain within TAZ that confers its complexing with USP1. Significantly, USP1 and TAZ were shown to co-occur in Triple-Negative breast cancer patients and can reduce relapse free survival. Our data has shown a novel role of USP1 in controlling TAZ protein levels that may give insight into new treatment options for triple negative breast cancer patients.**To request an accessible version of the file(s) associated with this item, contact [email protected]. Please include the item's persistent URL [http://hdl.handle.net/. . .] in your request.*
Foreign Economic Policies in the European Parliament
No full textPh.D.How do legislators form policy positions on comprehensive trade agreements and foreign aid? The European Parliament (EP) has become a veto player in the approval of EU trade agreements and foreign aid in the past decade. Yet, there is little empirical research on foreign economic policy positions in the EP. My dissertation consists of three studies on Members of the European Parliament's (MEPs) policy positions on trade and foreign aid. In Chapter 1, I provide a framework for understanding MEPs' policy positions on comprehensive trade agreements. Using an original dataset on trade voting records, I show that higher levels of public trust in government make MEPs more likely to vote in favor of comprehensive trade agreements, holding constant for economic and non-economic predictors. Chapter 2 analyzes mobilization of special interest groups on comprehensive trade agreements in the EP. I use data from the EU Transparency Register and elite surveys on MEPs' to show that frequent interactions with certain lobby groups predict MEPs' views toward the Transatlantic Trade and Investment Partnership (TTIP). In Chapter 3, I study the relationship between Foreign Direct Investment (FDI) into EU member states and MEPs' positions on trade policy and foreign aid. I collect data on five EP-votes concerning trade and financial aid to Ukraine. I argue and show that MEPs with higher levels of FDI from Russia in their electoral districts are more likely to vote against EU aid and trade with Ukraine. My three studies offer insights on MEPs' position-taking on trade policy and foreign aid. My findings improve scholars' ability to predict policy outcomes of EU foreign economic issues that have global economic and political ramifications.**To request an accessible version of the file(s) associated with this item, contact [email protected]. Please include the item's persistent URL [http://hdl.handle.net/. . .] in your request.*
Generalized WO-models of Principal Series Representations of SO(2n)
No full textPh.D.Let G be a a split even orthogonal group. We consider generalized Whittaker-Orthogonal functionals. These are defined by generalizing the WO-functionals of [3] which are equivariant with respect to the product of a unipotent group and a smaller odd orthogonal group. We give explicit criteria for a principal series representation of G to support such a functional.**To request an accessible version of the file(s) associated with this item, contact [email protected]. Please include the item's persistent URL [http://hdl.handle.net/. . .] in your request.*
Geometric Schur Duality over the Complex Numbers
No full textPh.D.In this paper we examine duality relationships of certain convolution algebras. Guillame Pouchin established a geometric Schur duality over finite fields via endomorphisms of function spaces equipped with a convolution product. This paper generalizes that work to function spaces over complex algebraic varieties. We define an integral constructed from the Euler characteristic and with this define a convolution product. First we define this integral and convolution and demonstrates some of their properties. Following that we prove a bicommutant theorem in our setting analogous to Pouchin's work. We then examine an application of this duality and study cases to extend the result. In particular we recover the classic Schur duality under specific conditions by working with flag varieties.**To request an accessible version of the file(s) associated with this item, contact [email protected]. Please include the item's persistent URL [http://hdl.handle.net/. . .] in your request.*
Striking Deals or Striking Down: How Different Size Winning Coalitions Respond to Violent and Nonviolent Protest
No full textPh.D.When governments face protest, a calculation must be made in determining how they will respond – either by ignoring, accommodating, repressing, or some combination of the three – to those protests. Governments must weigh the costs and benefits of the response they pursue when dealing with protesters. I draw on theories of the effectiveness of violent and nonviolent protest, as well as on theories surrounding the size of the winning coalition, to predict the likelihood the government responds in a given manner. I consider the nature of the protests as well as the size of the base of support government that is facing those protests and how this impacts the decision-making process. I focus in particular on the ability of the government to spread costs across the winning coalition, and the impact responses to a given type of protest have on perceptions of the government’s legitimacy. Looking further into those governments that rely on small winning coalitions to remain in power, I examine the size of the selectorate from which the coalition is drawn from to determine the impact of being able to more easily replace members of the winning coalition. I derive a series of hypotheses that are subsequently tested on protest events that occurred between 1990 and 2017. The empirical results indicate mixed support for the theoretical arguments put forth. In these findings, although the size of the winning coalition makes a difference in some scenarios, it is oftentimes the use of violence that drives the likelihood of a particular response to protesters, rather than the size of the winning coalition or selectorate. Additionally, although accommodation may be more likely when protesters are nonviolent, it is still not a response often received, regardless of the government characteristics. Violent protests appear effective in bringing about accommodation in some situations, but not without also being repressed. The empirical findings confirm a number of previous studies, as well as advance our understanding of government responses to protest by examining individual protest events and considering the size of the winning coalition and selectorate of the government.**To request an accessible version of the file(s) associated with this item, contact [email protected]. Please include the item's persistent URL [http://hdl.handle.net/. . .] in your request.*
Tailor-Made Materials: Inverse Engineering Compounds Using Feature Correlation
No full textM.S.In this thesis, we highlight a time efficient protocol to "Inverse Engineer" a preliminary set (1st generation library) of compounds, originating from a broad chemical space, to create novel molecules (2nd generation library). This 2nd generation library provides access to a more targetted set of compounds having dedicated and enhanced properties in a narrower chemical space. This technique allows us to control the target property of the 2nd generation library. We use the Genetic Algorithm, from our machine learning program package ChemML, in which the building blocks serve as "genes" for every "individual" molecule and each "individual" molecule is evaluated based on a fitness function we design. This protocol requires a data set consisting of compounds, along with their respective target properties, which serves as the 1st generation library. Using this 1st generation library, a sub-set consisting of the top 10 percent candidates, having the most desirable target property values, is selected as the "set of better performers". We identify the prevalent sub-structures that are over-expressed in this subset and distinguish the better performing sub-structures from the less promising ones. The extent of correlation of each feature, to the target property is understood using two, statistics-based, feature selection techniques: A-scores analysis and Z-score analysis (hyper geometric distribution analysis). We use the results from the statistical analysis to create moieties that serve as building blocks for the second generation library. We optimize the hyper-parameters of an Artificial Neural Network and fit the preliminary data set to the model, to serve as an evaluation function in the genetic algorithm, to predict and evaluate the novel molecules created. We explore the utilities of our results from the feature correlation to highlight the over-expressed features and also showcase the results of this protocol to generate a second generation library, with improved target property values. This closes the gap between the shortfall of, a single, conventionally designed material having all distinct properties for a specific application, and the industry's demand for an intricately designed material having desired set of properties. This protocol supersedes the conventional techniques of fabricating novel materials because it is possible to predict the qualities of the end product before hand. Unlike the typical frameworks, the results from this protocol yield an array of options, which serves as a well defined chemical space, consisting of all the potential compounds and their respective target properties, which fulfill the necessary criteria for it's application in the industry.**To request an accessible version of the file(s) associated with this item, contact [email protected]. Please include the item's persistent URL [http://hdl.handle.net/. . .] in your request.*
"The Mind Sneezing": Modernist Poetry and the Para-Mediation of Humor
No full textPh.D.This dissertation re-figures humorous gestures obscured by the perceived seriousness of modernist poetry. Perhaps because the Victorian legacy of "high seriousness" was most invested in poetry, most studies of modernist humor have been aimed at fiction and film. Meanwhile, the field of humor studies has largely avoided theorizing what most consider its key counter-concept, as if defining seriousness were unnecessary or impossible. I argue that humor and seriousness are, in the terms of German media philosophy, cultural techniques of distinction; humor and seriousness are among the primary tools humans use to define categories—especially that of the human—and to stake out and maintain values, spaces, and discourses. Because most media concepts are products of serious techniques, we need a concept of para-mediation to adequately account for humor. I take humor to be most fundamentally rooted in the "friction" between mind and body that leads the British modernist Wyndham Lewis to define laughter as "the mind sneezing." Chapter One outlines my interventions in modernist literary studies, humor studies, and media theory: complicating the critical construction of modernist seriousness, identifying seriousness as the blind spot of humor studies, and facilitating the cross-pollination of theories of humor and theories of mediation. I develop these insights through three case studies in the nascent poetics of American modernism: Stephen Crane's affectation and "manual wit" in The Black Riders and other lines (1895), Gertrude Stein's deliberate "confusion" of seriousness and humor in Tender Buttons (1914), and Marianne Moore's witty “sleights of mind” in Observations (1924). This project explores humor's potential to transform media-theoretical approaches to literature while illustrating some of the ways that American modernist poets subverted or rewrote the rules regarding humor's place in poetry.**To request an accessible version of the file(s) associated with this item, contact [email protected]. Please include the item's persistent URL [http://hdl.handle.net/. . .] in your request.*
Groomed Event Shapes
No full textPh.D.The Higgs boson was the last fundamental particle discovered by the LHC back in 2012. Since then, no signatures of new physics has been detected. As such, we are entering the precision era of the LHC. This thesis will begin with an introduction of the Standard Model of particle physics, with a specific focus on QCD interactions. An overview of jets, focusing on why jets are useful in studying QCD, as well as how they are defined and how to groom them to study their substructure. Event shapes are then introduced as tools to tune Monte Carlo generators, but more importantly are used to find robust, precise values of αs. Grooming techniques, originally designed for jets formed at the LHC, are applied to event shapes at e+e− colliders at LEP scales, as well as pp colliders at LHC scales. Results are provided indicating how groomed event shapes may improve the precision of αs extractions from LEP e+e− colliders, as well as providing an incentive to study event shapes at the LHC.**To request an accessible version of the file(s) associated with this item, contact [email protected]. Please include the item's persistent URL [http://hdl.handle.net/. . .] in your request.*
Predicting Combined Sewer Overflow Occurrence from Rainfall Data
No full textM.S.The occurrence of combined sewer overflows (CSOs) is one of the most pressing environmental issues facing many cities with a combined sewer system (CSS). Combined sewers often exceed the capacity of the CSS or wastewater treatment plant and discharged directly into water bodies without treatment. CSO discharges contain a variety of contaminants and degrade receiving water quality. To manage CSOs, it is necessary to characterize their occurrence, which generally requires monitoring CSOs for long periods or building complex hydrological models for simulation. However, monitoring CSOs is very expensive, and hydrological models are often too time-consuming to predict CSO performance in real-time. Thus, this work aims to develop a simpler method for predicting CSO occurrences using rainfall characteristics and investigate which rainfall characteristic(s) can be the best predictor. The city of Buffalo, New York, is taken as a case study. A calibrated hydrological model was used for one-year continuous simulation to generate CSO discharges at 52 CSO outfalls. An R-language code was developed to analyze the characteristics of rainfall and CSO events. Rainfall characteristic quantities such as duration of the rainfall event, rainfall depth, and maximum rainfall intensity were assessed for their prediction power of CSO occurrence. Prediction accuracy with single characteristic, rainfall depth, or maximum rainfall intensity, ranged from 80 to 100%, while rainfall duration was found not a good predictor. Rainfall depth was found to be a better predictor for sewersheds with larger areas and smaller imperviousness percent, while maximum rainfall intensity was found better for sewersheds with smaller areas and bigger imperviousness percent. Moreover, the decision tree algorithm was utilized to combine more than one rainfall characteristic. The average prediction accuracy was slightly improved from 93% (using a single characteristic) to 95%. These results reveal that rainfall data can yield accurate prediction of CSO occurrence, and the proposed method can be an effective alternative to complex hydrological models and/or expensive monitor for managing CSOs.**To request an accessible version of the file(s) associated with this item, contact [email protected]. Please include the item's persistent URL [http://hdl.handle.net/. . .] in your request.*
Representation Learning for Treatment Effect Estimation
No full textPh.D.Treatment Effect is defined as the difference between the outcomes of different treatments, following the potential outcome framework, where the treatment is the action that exposes (or subjects) to a unit which can be any physical object, and the outcome is the result after performing the treatment. The estimated effect facilitates decision making across various domains, such as healthcare, business, and sociology science. Due to the wide applications, treatment effect estimation is an attractive research topic for a long time. The treatment effect can be estimated from randomized controlled trials (RCTs) or observational data. Because RCTs can be expensive, time-consuming, and sometimes unethical, nowadays, estimating the effect from the observational data has become a trending topic due to its large amount of availability. Estimating the treatment effect from observational data comes across two challenges. The first one is that we can only observe the factual outcome and unable to observe the counterfactual outcome, which is the result if the unit has chosen a different treatment option. The second one is that treatments are typically not assigned at random in observational data, which may result in the treated population differs significantly from the general population. The flourishing of machine learning brings new vitality to solve the above two challenges and provides a more accurate estimation of the treatment effect. Towards the above challenges, we first propose the similarity-preserved representation learning framework for counterfactual outcome estimation. It learns a balanced representation across different treatment groups with pairwise similarity information adaptively preserved so that it can overcome the selection bias and meanwhile improve the estimation quality. Then we improve the traditional matching-based treatment effect estimation algorithm based on the fact the including the near-instrumental variables in the matching procedure would amplify the estimation bias. It expands the scope of application of treatment effect estimation to the data containing text type covariates. The proposed matching algorithm performs nearest neighbor matching upon the latent space with information related to near-instrumental variables being filtered out. Machine learning methods enhance the development of causal inference; meanwhile, causal inference also helps machine learning methods. The simple pursuit of predictive accuracy is insufficient for modern machine learning research, and correctness and interpretability are also the targets of machine learning methods. In the last part, we will investigate the applications of treatment effect estimation in trustworthy machine learning, and in this thesis, we focus on model interpretation. We propose to utilize the treatment effect estimation methods to detect the important components in model prediction. In this thesis, we develop various novel representation methods to improve the treatment effect estimation. Meanwhile, we also explore the potential applications of treatment effect estimation in the machine learning area. With the development of treatment effect estimation and the machine learning area, there are still numerous challenges and opportunities in combining the two areas so that they can enhance each other.**To request an accessible version of the file(s) associated with this item, contact [email protected]. Please include the item's persistent URL [http://hdl.handle.net/. . .] in your request.*