Carleton University Institutional Repository
Not a member yet
20816 research outputs found
Sort by
Bacterial Stress Management: How Pseudomonas Aeruginosa Survives Reactive Chlorine Species
No full textPseudomonas aeruginosa is a highly pathogenic Gram-negative bacterium due to its resistance to many commonly used disinfectants and antibiotics, the production of numerous virulence factors, and the formation of robust biofilms. In their natural habitats, bacteria are constantly confronted with environmental stressors, including oxidants like reactive chlorine species (RCS). The most widely used RCS is sodium hypochlorite (NaOCl; household bleach) and its active ingredient, hypochlorous acid (HOCl). Besides its function as a strong disinfectant, HOCl is also produced by the human innate immune system as a defense against invading pathogens. Despite the potent antimicrobial effect of NaOCl and HOCl, bacteria have developed sophisticated mechanisms to adapt to the damages caused by these disinfectants, many of which remain largely unknown. Therefore, this study aimed to identify and characterize genes involved in the resistance of P. aeruginosa planktonic and biofilm cells to NaOCl by combining high-throughput mutant library screenings and targeted mutant analyses. In our screening of planktonic bacteria, we identified ten genes (nrdJa, bvlR, hcnA, orn, sucC, cysZ, nuoJ, PA4166, opmQ, and thiC) whose absence increased the susceptibility of P. aeruginosa to NaOCl. Follow-up analyses showed that hydrogen cyanide (HCN) increases the resistance of P. aeruginosa to NaOCl by acting as a scavenging molecule. Moreover, another ten genes (i.e., apaH, PA0793, acsA, PA1506, PA1547, PA3728, yajC, queA, PA3869, or PA14_32840) were identified in our biofilm screening, in which their absence increased susceptibility of respective P. aeruginosa biofilms to NaOCl by 4-fold compared to the PA14 wild-type strain. Finally, we showed that the biofilm matrix polysaccharides Psl and Pel protect against oxidative stressors since their absence significantly increased the susceptibility of P. aeruginosa biofilms to NaOCl and H2O2. Further analyses revealed that Pel is more important for oxidative stress resistance in P. aeruginosa biofilms than its Psl counterpart. Understanding these resistance mechanisms will help to understand the adaptation of pathogens to RCS and their survival strategies in the host and the environment, which will be useful to improve the control of bacterial growth and biofilms
Methods for Classifying Driver Engagement in Autonomous Vehicles Using Physiological Sensors
No full textAccurately measuring driver attention and tracking eye gaze in dynamic environments is crucial for advancing global road safety and the adoption of autonomous vehicles. Physiological sensors and machine learning models are employed to classify driver attention during real world autonomous and manual driving scenarios, demonstrating high accuracy identifying attentive drivers. Traditional gaze measurement techniques struggle with dynamic contexts, but advancements in deep learning offer promising solutions. This work introduces methods leveraging convolutional neural networks and transfer learning to enhance the segmentation of a driver's field of view into relevant areas, successfully segmenting thousands of video frames automatically. The automatic segmentation of video frames enabled the creation of novel gaze metrics. These innovations in gaze metrics further enhance the classification of driver attention with physiological metrics. Attention monitoring systems are crucial for ensuring safety and effective handover in semi-autonomous vehicles, offering substantial contributions to driver safety and attentiveness assessment methodologies
Basic Psychological Need Satisfaction and Frustration During Social Media Use and Mental Health Among University Students
No full textResearch on psychological implications of social media use (SMU) is inconclusive. Much of the research focuses on SMU quantity and has given little attention to subjective experiences on SMU. The goal of this study was to explore if the link between SMU screen time and mental health depends on experiences of basic psychological need satisfaction and frustration during SMU. Using a randomized control trial, 100 undergraduate students were assigned to either an experimental (i.e., limit SMU to 60 mins/day) or control (i.e., no SMU time limit) group. SMU screen time was collected for 4 weeks using smartphone-generated reports. Positive and negative affect, depression, anxiety, and need experiences during SMU were assessed at baseline and post-intervention. A favourable effect of reducing SMU was observed on positive affect and depression in users with above average SMU-based need frustration. Therefore, reducing SMU may be most beneficial to users with highly need frustrating SMU
Diagnosis and Prognosis of Lithium-Ion Batteries State Using Machine Learning
No full textLithium-ion batteries (LIBs) have emerged as the primary energy storage solution, yet concerns over cost, safety, and reliability hinder their widespread use, notably in electric vehicles (EVs). Battery aging causes capacity and power decline, impacting EVs' range and safety, urging the need for sustainable energy solutions. Thus, precise health estimation and temperature monitoring are essential for prolonging battery life and minimizing risks. This thesis investigates methods to enhance LIBs' longevity and performance through accurate health and temperature estimation. Leveraging nonlinear autoregressive with external input recurrent neural networks (NARX-RNN), the study assesses their effectiveness in state of health (SOH) estimation and remaining useful life (RUL) prediction, achieving impressive performance for online applications, especially in EVs. However, precise health estimation faces challenges, particularly in inconsistent industrial settings. A novel SOH estimation model is introduced, employing NARX-RNN without imputation techniques, showcasing resilience across various datasets and chemistries, suggesting broad industrial applicability. Temperature management is critical for LIBs' performance and safety. To address concerns over temperature sensor costs and reliability, a hybrid method is proposed for accurate surface temperature (ST) estimation, integrating convolutional neural networks (CNN), long-short term memory (LSTM), and deep neural networks (DNN), promising enhanced safety and performance in practical LIB applications
Arsenic Mobilization from Thawing Permafrost
No full textThis document is the unedited Author’s version of a Submitted Work that was subsequently accepted for publication in ACS Earth and Space Chemistry, copyright © 2024 the Authors. To access the final edited and published work see : https://pubs.acs.org/doi/full/10.1021/acsearthspacechem.3c00355Thawing permafrost releases labile organic carbon and alters groundwater geochemistry and hydrology with uncertain outcomes for the mobility of hazardous metal(loid)s. Managing water quality in thawing permafrost regions is predicated on a detailed understanding of the speciation and abundance of metal(loid)s in permafrost soils, and porewaters produced during thaw, which remains limited at present. This study contributes new knowledge on the sources and fate of arsenic during thaw of organic-rich permafrost soil using samples collected from a subarctic permafrost region associated with geogenic arsenic (Dawson Range, Yukon, Canada). Several permafrost cores and active-layer samples from this region were analyzed for their solid-phase and fluid geochemical characteristics and their arsenic speciation. Porewaters were extracted from permafrost cores thawed under anaerobic conditions for aqueous geochemical analyses. Bedrock samples from the field site were also analyzed for their arsenic speciation. X-ray diffraction and X-ray near-edge spectroscopy (XANES) analyses of weathered bedrock upgradient of soil sampling locations contains arsenic(V) hosted in iron-(oxyhydr)oxides and scorodite. XANES and micro X-ray fluorescence analyses of permafrost soils indicate a mixture of arsenic(III) and arsenic(V), indicating redox recycling of arsenic. Soil-bound arsenic is co-located with iron, likely as arseniferous iron-(oxyhydr)oxides that have been encapsulated by aggrading permafrost over geologic time. However, permafrost thaw produced porewater containing elevated dissolved arsenic (median 40 μg L–1, range 2–96 μg L–1). Thawed permafrost porewater also contained elevated dissolved iron (median 5.5 mg L–1, range 0.5–40 mg L–1) and dissolved organic carbon (median 423 mg L–1, range 72–3,240 mg L–1), indicative of reducing conditions. This study highlights that arsenic can be found in reactive forms in permafrost soil. Thaw of this material can release arsenic to porewater leading to poor water quality
Genome-wide analysis contributes to and promotes adaptation of the soybean, Glycine max, to Canadian agricultural landscapes
No full textThe soybean, Glycine max (L.) Merr., is an important crop due to its contribution to human diet, animal feed, and sustainable crop rotation but production is limited by photoperiod and damaging stressors. Research is carried out in the continuum between applied and functional genomics with cutting edge techniques, contributing information to the scientific community. Germination data from cultivars subjected to cold imbibition was analyzed in a genome-wide associate study (GWAS), revealing three quantitative trait loci (QTL) controlling tolerance to cold imbibition. Latent phenotypes of agronomic traits were developed and used in GWAS and epistatic analysis, revealing novel QTL associated with maturity, yield, protein, and oil. This demonstrates that valuable QTL can be detected when information like genotype by environment interaction is included. Techniques bridging the gap between applied and functional genomics, identifying novel genes and their functions by gleaning information from multi-disciplinary sources and model organisms, will facilitate crop improvement
Analytical Studies of Atmospheric Gravity Waves and Convection under the Anelastic Approximation
No full textThis thesis provide a mathematical model that represents internal gravity waves and convection caused by heat forcing in the atmosphere. The purpose is to investigate the mechanisms that generate internal gravity waves in the lower at- mosphere due to deep heating and convection. We analyze a two-dimensional one-layer atmospheric model with stable stratification and an unsteady convec- tive layer. The governing equations are based on the equations for fluid mass, momentum, and energy conservation. A nonhomogeneous term in the energy conservation equation represents thermal force. We investigate several configu- rations based on the vertical structure and depth of the thermal forcing. First, we derive exact analytical solutions for the linearized equations where the ampli- tude of the perturbations is time independent. Then, in each layer, we look at the case where the magnitude of the perturbations varies with time and obtain approximate time-dependent solutions. In the limit of infinite time, the linear gravity wave solution approaches the steady solution. We also investigate the nonlinear gravity wave problem, conduct a weakly-nonlinear analysis, and inves- tigate the evolution of the mean flow with time as a result of the waves’ nonlinear interactions. Suggested next approaches for this study could involve incorporating numeri- cal simulations to explore the case under investigation and further analyzing the convection layer. In addition, the solutions presented here may be used as a starting point for a convective gravity wave drag parameterization approach that could be applied in large-scale atmospheric general circulation models
Six Acres: The Residence & Office of Edmonton Architectural Partnership, Wallbridge & Imrie
No full textThis thesis investigates the architectural and historical significance of Six Acres, the residence and office of Wallbridge & Imrie, Architects in Edmonton, Alberta. Through a detailed examination of this unique property, it addresses broader themes of domesticity, house design, and the role of women in architecture, while situating Six Acres within the mid-20th century architectural landscape of Alberta. Six Acres exemplifies Wallbridge & Imrie's design philosophy, merging living and working spaces in a way that reflects their personal and professional values. The study aims to document a complete history of Six Acres, serving as a comprehensive guide and illuminating often overlooked aspects of architectural history. Structured into five key sections—The Architects, The Building, The Property, The 'Afterlife,' The Afterword—the thesis provides a thorough analysis of Six Acres and its impact on architectural discourse. It challenges conventional narratives and contributes to a nuanced understanding of mid-century modernism in Alberta, Canada
Government of Canada Social Media Monitoring and Its Role in Public Environment Analysis
No full textOver centuries, governments have employed various mechanisms to measure and understand their populations. As information and computer technologies progressed in the 20th century, governments adopted new ways of communicating with citizens and gathering statistics to inform decision making; at the same time, digital platforms such as social media became increasingly integral to social life. This dissertation examines the Government of Canada (GC) and its use of social media monitoring to understand the public environment, specifically within the context of communications work. It situates this analysis within a broader context of public administration paradigms and in the affordances of social media monitoring tools, in order to understand how this monitoring constructs a particular understanding of the public. The dissertation employed interpretive content analysis to conduct primary research, with data consisting of expert interviews, survey results and GC documents and policies. Researcher positionality was also a core aspect of analysis; prior knowledge and experience of social media monitoring inspired the subject of study and informed research questions. It equally afforded the researcher access to the 71 participants interviewed in the study, and the opportunity to share research findings with the Privy Council Office. This research found that participating GC departments often sought to use social media monitoring to support existing work done in communications branches, and in many ways approached monitoring with aims that align with tenets of Digital Era Governance. At the same time, existing conditions within the GC bore marks of New Public Management-style governance that limited the GC’s capacity to undertake monitoring in a methodologically sound and critically engaged way. Communications branches were not necessarily equipped to understand and address the challenges of using social media data, particularly with regards to privacy requirements. They also depended heavily on tool vendors for the technologies and training, which facilitated the adoption of dataist rhetoric already being espoused by public and private sectors alike. Ultimately, the public environment analysis conducted by GC communications branches through social media monitoring is partial and biased, while largely contravening privacy requirements. This is particularly significant as departments look to expand monitoring and its applications in government
Application of Machine Learning and Deep Learning Approaches Cheminformatic for Drug Discovery
No full textThis study evaluates molecular property prediction (MPP) using machine learning (ML) and deep learning (DL) approaches on Morgan fingerprints. ML methods such as Random Forest, XGBoost, and SVM were compared, with Morgan fingerprints showing improved performance. DL techniques included a pre-trained bi-directional encoder (BERT) for semantic representation extraction and LSTM-based vectorization using mol2vec. A hybrid model combining seq2seq LSTM with XGBoost achieved high accuracy in property prediction, leveraging neural networks' nonlinear generalization. This methodology demonstrates the efficacy of combining ML and DL techniques for accurate molecular property prediction. This innovative approach integrates diverse elements, creating a nonlinear network. Across ML and DL classifiers, active and inactive compounds were predicted with 0.97 accuracy. F-1 scores for Random Forest, XGBoost, and SVM are 0.96, 0.97 and 0.96, respectively, indicating robust model performance