Western University Open Repository
Not a member yet
30251 research outputs found
Sort by
Molecular characterization of tomato brown rugose fruit virus-host interactions and development of biocontrol strategies
No full textTomato brown rugose fruit virus (ToBRFV, Tobamovirus fructirugosum) is an emerging RNA virus that threatens global tomato production. The virus can break down durable extreme resistance (ER) to its related tobamoviruses conferred by the NLR gene Tm-22. Currently, no effective control strategies are available. This thesis research was designed to better understand ToBRFV at the molecular level, explore the mechanisms by which ToBRFV overcomes Tm-22-mediated resistance, and develop a disease management approach through biocontrol.
In this dissertation, the genomes of three ToBRFV Canadian isolates were cloned and sequenced, and full-length cDNA infectious clones were constructed. Defective mutation analyses revealed that ToBRFV movement protein (MP) and coat protein (CP) contribute to ToBRFV local accumulation and are critical for systemic infectivity. Two highly conserved CP residues, D89 and R114, are essential for ToBRFV long-distance movement. Alanine substitutions of these two residues disrupted CP self-interaction and virion formation. The 126 kDa replicase was identified as the viral RNA silencing suppressor.
Although ToBRFV has evolved the ability to evade Tm-22-mediated ER, graded inoculation assays indicated that it still confers attenuated, dose-dependent resistance in both Tm-22-carrying Nicotiana benthamiana and tomato plants. This novel discovery ascertains the value of Tm-22 in resistance breeding programs. Mechanistic studies demonstrated that the SGT1 (suppressor of G-two allele of Skp1)/HSP90 (heat shock protein 90) complex, a core complex known to be required for innate immunity in plants, interacts with ToBRFV MP, which is responsible for breakdown of Tm-22 resistance. This interaction boosts ubiquitination and facilitates proteasome degradation via SCFFBS1, a prototypical Cullin-RING E3 ubiquitin ligase complex. Phospho-mimetic assays further revealed that phosphorylated SGT1 and HSP90 negatively regulate MP abundance and ToBRFV accumulation. Application of exogenous salicylic acid (SA) could activate SGT1 and HSP90 through stabilizing the master regulator NPR1 (nonexpressor of pathogenesis-related genes 1). As a counteracting mechanism, ToBRFV CP antagonizes host defenses by specifically targeting the SGT1-HSP90 complex for proteasomal degradation.
To develop a biocontrol agent through cross-protection, a number of ToBRFV mutants were generated and screened. Three and four single amino acid mutants substantially suppressed ToBRFV symptom development and viral accumulation in N. benthamiana and tomato, respectively. Among them, the strong protection efficacy of the S643F mutant was verified in a greenhouse trial under growth conditions that mimic commercial greenhouse production.Shaokang Zhang, 202
Experiences of loneliness and their relation to perceived social support, family functioning, and mental health among Canadian Veterans
No full textIntroduction: Veterans are particularly susceptible to heightened levels of social isolation and loneliness. This study aimed
to explore the relations between loneliness, perceived social support, and mental health symptoms in Canadian Armed
Forces (CAF) Veterans. Methods: Data were drawn from a national online survey of Canadian Veterans and Veteran
spouses. Self-report measures assessed demographic and service-related factors, loneliness, perceived social support, family
functioning, positive mental health, and symptoms of posttraumatic stress disorder (PTSD), depression, anxiety, and
moral injury. Because data collection took place during the COVID-19 pandemic, certain context-specific variables were
also collected, including exposure to COVID-19 and pandemic-related behavioural changes. Multiple linear regressions
were used to construct explanatory models of loneliness. Results: Adjusting for other demographic variables, higher levels
of loneliness in Veterans were associated with lower perceived social support, difficulties in family functioning, elevated
mental health symptoms (i.e., PTSD, depression, and anxiety), less contact outside one’s home, and those who were single/
unmarried. Conversely, positive mental health was associated with reduced feelings of loneliness. Discussion: The degree
of loneliness is linked to several mental health, social, and demographic factors, which could serve as key indicators for
identifying at-risk individuals or act as focal points for intervention. Implementing strategies that promote positive mental
health and foster social connections may offer a promising approach to reducing the impact of loneliness
Whose Justice? Promoting a Holistic and Inclusive Conception of Gender Justice in Transitional Societies
No full textThe principal goal of this thesis is to determine the normative content of the term 'gender justice' in transitional justice (TJ) and international criminal law (ICL). It maps and outlines the common content - and therefore a definition - of gender justice through literature reviews and semi-structured interviews with gender justice experts, including those directly working with victims/survivors of gender-based violence, and finds that gender equality and non-discrimination are the foundational legal bases of this term. The thesis argues that the term 'gender justice' ought to come into wider legal use with this clear definition because it can helpfully express, in one phrase, the complex feminist meaning of gender-equal and non-discriminatory justice, one that benefits victims/survivors of gender-based violence in TJ and ICL scenarios. The thesis applies the proposed definition to two examples of TJ mechanisms: the International Criminal Court’s (ICC’s) complementarity system and the operation of Liberian peace huts (peace huts). In doing so, it describes and explains how the application of this definition and its embedded tenets and practices help to reveal how the ICC and the peace huts can better support gender equality and non-discrimination than without such a definition. It also argues that gender justice must be strategically pursued, holistic, and responsive to varying forms of injustice beyond criminal harms to avoid achieving partial gender justice. In sum, this thesis advances the TJ and ICL fields by proposing a clear definition of gender justice that, applied consistently, could improve existing and future justice practices to be more fully gender-equal and non-discriminatory, and therefore in line with fundamental norms of international law.Loyce Mrewa, 202
Trustworthy AI Under Distribution Shifts: Enhancing Generalization and Fairness in Deep Learning
No full textRecent advances in machine learning and deep learning have achieved remarkable success across a wide range of applications, yet they largely rely on data-driven paradigms that assume the training and test data share similar distributions. In real-world scenarios, however, this assumption often fails. Distribution shifts between training and deployment environments or among population subgroups can significantly compromise the performance and reliability of AI systems. Moreover, privacy constraints frequently prevent direct access to source data, and ethical concerns surrounding fairness raise additional challenges for the trustworthy use of AI in sensitive domains such as healthcare and finance. These factors together highlight the urgent need for a deeper and more principled understanding of trustworthy AI under distribution shifts.
This thesis examines the influence of distribution shifts along two complementary dimensions of trustworthy AI: generalizability and fairness. On one hand, we focus on the noisy and unreliable signals caused by distribution shift, which place learning in a weak-supervision regime and thereby limit AI systems’ generalizability. On the other hand, we focus on the distribution shifts that arise from structural data imbalance or scarcity rooted in historical and societal factors, which give rise to socially unfavorable model behaviors and, in turn, exacerbate fairness issues in the decision-making process.
From the generalizability perspective, the thesis studies the domain adaptation problem under data privacy constraints, formulated as the Source-Free Domain Adaptation (SFDA) problem. SFDA is examined in both classification and regression tasks, with a focus on pseudo-label noise mitigation and uncertainty modeling, respectively. For the classification problem, a noise-robust loss–based noise-and-variance control method (NVC-LLN) is proposed to mitigate the unbounded label noise inherent in SFDA. For the regression task, a histogram-based uncertainty reconstruction framework, MERCI, is proposed to refine continuous supervision signals and align feature presentations under distribution shift.
For the fairness problem, this thesis focuses on the multiple-subgroup situation, ranging from a single sensitive attribute with multiple subgroups to multiple sensitive attributes with intersectional bias, covering both unfairness discovery and mitigation. For subgroup discovery, a Bias-Guided Generative Network (BGGN) is proposed to identify underrepresented subgroups instead of relying on traditional search methods. For mitigation, a generalization error bound is derived under the PAC-Bayes framework, motivating a probabilistic-predictor-based bilevel optimization algorithm, FAMS, which achieves both fairness and accuracy across multiple subgroups under limited-data conditions.
Collectively, these studies advance the development of responsible, adaptive, and human-centric AI systems. These systems are capable of performing reliably across diverse and dynamic environments.Gezheng Xu, 202
Undoing Erasure: A Comparative Study Of Trauma And Testimony In Arabic And Asian Literature
No full textThis study examines how literary works grapple with the challenges of representing trauma, memory, and testimony, focusing on three primary texts: Hoda Barakat’s Voices of the Lost, Haifa Zangana’s Dreaming of Baghdad, and Nora Okja Keller’s Comfort Woman. Each of these novels— whether in epistolary form, autobiographical testimony, and intergenerational fiction—explores the complex interplay between personal suffering and cultural silencing in the wake of widespread violence. Barakat’s novel, set against a nameless conflict-ravaged landscape that evokes Arab countries, depicts characters whose undelivered letters reveal profound isolation and the breakdown of communication following displacement. Zangana’s memoir describes the experiences of political imprisonment and torture under Saddam Hussein’s regime, illustrating how narratives of incarceration challenge both official history and patriarchal norms. Keller’s intergenerational story draws attention to the historical trauma of “comfort women”, exposing the haunting legacy of wartime sexual enslavement passed down through mother-daughter relationships. Throughout these texts, the authors employ non-linear structures, shifting timeframes, and narrative fragmentation—devices common to trauma literature—to portray the belated, recursive nature of traumatic recollection. They underscore the ethical dilemma of bearing witness: survivors must recount experiences that resist full expression, even as they encounter cultural and political forces that prefer denial or erasure. This study argues that by voicing private anguish, the works under consideration also act as collective testimony, confronting official narratives and preserving marginalized memories. In highlighting the polyvocal scope of trauma—ranging from political conflicts to wartime atrocities—this analysis ultimately illustrates the border-crossing nature of suffering and the vital role of literature as an archive for the dispossessed.
Keywords: trauma, testimony, memory, war literature, displacement, women’s narratives, epistolary fiction, political imprisonment, silence
Exploring the Influence of Culture and Religion on Nutrition and Physical Activity Behaviours of Pregnant Arab Muslim Individuals
No full textGestational diabetes mellitus (GDM) is a growing concern globally and in Canada. Given that there are many high-risk groups, such as Arab individuals, it is important to ensure that GDM prevention interventions are available that fit population needs. Therefore, this qualitative descriptive study explored the influence of culture and religion on nutrition and physical activity behaviours in pregnant Arab Muslim individuals. Thirty-two pregnant or <1-year post-partum individuals engaged in semi-structured interviews. Summative content and thematic analysis were used to understand the influence of cultural and religion on these behaviours. Four core themes emerged providing a deeper understanding: 1) Importance of food in family and community; 2) Physical activity is integrated into everyday life; 3) Thinking about the baby; and 4) Factors involved in lifestyle choices. Information gathered may help guide the modification of current nutrition and physical activity lifestyle interventions for this specific at risk population.Shubhika Mahakul, 202
Validation of a Thermal Resistance Model and Thermal Behavior of an Embedded Cylindrical Copper Heat Pipe in a P20 Tool Steel Cylinder for Industrial Heating Applications
No full textThis thesis validates a heat pipe Thermal Resistance Model (TRM) implemented as a dynamic boundary condition in a Computational Fluid Dynamics (CFD) simulation to study the transient thermal behaviour of a P20 steel cylinder with a centrally embedded copper–water heat pipe. Experimental tests were carried out using an instrumented P20 steel apparatus to measure the axial and radial temperature evolution during controlled heating. Custom User-Defined Functions (UDFs) that dynamically classified heat-pipe wall sections, calculated local resistances, and applied a time-dependent heat-flux profile were used to integrate the TRM, which was based on pool-boiling, film-boiling, and condensation correlations, into Ansys Fluent. With variations within the thermocouple uncertainty at the heat pipe's mid-height and condenser sites, numerical temperature fields closely matched experimental results, proving that the implemented TRM offers an effective computational substitute for explicit multiphase modelling.Camila Solano Vergara, 202
The Lived Experiences of Women Para Ice Hockey Players
No full textIn Canada the expectation is that men and women should have equal access opportunity to sport
participation, including our national pastime of hockey. However, in the co-ed environment of
para ice hockey this is not the case. This thesis examines why there are fewer women participating
in para ice hockey than men; by using feminist disability theory with qualitative methods and an
intersectional focus on gender and disability. Data was collected from: observations over the 2023
2024 season, interviews with women players, and from the writers’ experience with the sport as a
nondisabled woman coach/trainer. Analysis of the data showed that the women attribute overall
enjoyment of the sport to shared experiences of disability which creates an inclusive environment.
However, there are still moments when women must prove they belong in the sport. These findings
warrant the need for education and training to improve the sports’ culture for gender inclusion.Kaitlyn Guernsey, 202
Climate-Informed Design Flood Estimation in a Nonstationary Climate
No full textThe increasing magnitude and frequency of high flows driven by extreme precipitation, snowmelt, and other hydroclimatic processes under a changing climate pose growing challenges for the design, operation, and safety of dams and levees. Traditional design flood estimation methods, including frequency-based approaches and the use of Probable Maximum Floods (PMF) derived from Probable Maximum Precipitation (PMP), often assume climatic stationarity and simplified thermodynamic scaling that emphasize moisture availability while neglecting dynamic processes. This dissertation presents a comprehensive investigation into climate-resilient, physically and statistically grounded, and uncertainty-informed flood estimation methods integrating atmospheric drivers and climate model ensemble analysis. The first study critically reviews Canadian and international design flood practices, highlighting limitations in addressing regional vulnerability, nonstationarity, and the physical drivers of extreme events. A resilience-based framework is proposed, grounded in four key pillars: threshold, coping, recovery, and adaptive capacity. Building on this foundation, a machine learning-based regional flood frequency analysis framework is developed that incorporates large ensembles of climate simulations to project future spring flood quantiles in Southern Ontario. Results show spatially variable changes ranging from -13% to +20% under different warming levels. Anthropogenic Climate Change (ACC) and Internal Climate Variability (ICV) responses intensify with warming, with ICV dominating across most catchments, while ACC influence is more evident in western regions. We then investigate the thermodynamic and dynamic drivers of PMP-driven extreme precipitation across North America. Using reanalysis data and ground observations from over 6,000 stations, it demonstrates that moisture convergence and atmospheric instability are often more influential than moisture availability alone. The final study develops a new framework that integrates large climate ensembles, spatial storm transposition, and pseudo-observational benchmarking to reduce sampling uncertainty and evaluate systematic biases in PMP return levels under different warming scenarios. Results show that ensemble pooling without spatial transposition is insufficient to robustly constrain the tail behavior of precipitation distributions at extremely low exceedance probabilities. Furthermore, PMP return levels are projected to increase by up to 77% under 3°C warming, across 2730 dam locations in Northeastern North America. Together, these studies provide an integrated advancement in design flood estimation under climate change
Data-driven machine learning for Cas9 activity prediction
No full textThe CRISPR-Cas9 system has seen widespread use across a multitude of applications from genome editing to its use as a strain-specific antimicrobial. In spite of its potential, on-target activity varies depending upon the DNA target site selected. Given this issue, the ability to predict activity has become an essential component for practical use. While substantial progress has been made regarding the development of predictive models for use in eukaryotes, few models exist for bacterial application, and notably, predictions from eukaryotic models do not transfer to bacteria. Therefore, the central focus of this work was to develop techniques for the accurate prediction of bacterial sgRNA/Cas9 activity using machine learning, creating a set of models referred to as crisprHAL. Using these models we demonstrate that predictive performance is driven primarily by model-adjacent strategies, rather than specific improvements to the deep learning architecture. The initial models facilitated a single goal: transfer learning with small high-quality datasets, showing the importance of data quality as it relates to performance. This focus on non-architectural improvement continued throughout the work, emphasizing accuracy in activity scores, the inclusion of biologically relevant adjacent nucleotides, and ensemble methods which capture uncertainty in predictions. Beyond activity prediction, the methods presented provide a framework for biological inference. Using model performance as a guide, analyses revealed conserved target site adjacent preferences as long range effects which impact activity. Notably, for the SaCas9 nuclease, we uncovered a critical PAM-adjacent position and methylation-dependent suppression of activity at GATC motifs. Combined, this illustrates how machine learning can be used for more than just predictions, but as a lens for discovery. This work culminates in an uncertainty-aware framework which combines data quality improvements with the capture of variance in activity scores and model training. Together, these models attain strong performance across datasets, assay designs, nuclease variants, and bacterial species — a capability distinctly absent from eukaryotic models. With this work, we provide not just an accurate set of tools for researchers, but a set of methods by which to explore the biology underlying the CRISPR-Cas9 system