27716 research outputs found
Sort by
Transferable representation learning for drug discovery
Get PDFDrug discovery seeks to identify new candidate medications that can effectively treat human diseases with acceptable developability. Traditional computational and machine learning methods leverage handcrafted domain features for drug screening but suffer from poor transferability due to the vast chemical search space. In this thesis, we introduce transferable representation learning, a prominent approach within deep learning, to address different domain transferability challenges in drug discovery. Specifically, we develop three deep learning-based frameworks to learn transferable representations that adapt to key drug-related tasks, improving specific transferability for drug-target interaction prediction and enhancing generic transferability for molecular property prediction and inverse protein folding.
For drug-target interaction prediction, we first propose a low-bias evaluation strategy to effectively validate specific transferability. After that, we develop a bilinear attention network-based framework incorporating domain adaptation to improve performance under both in-domain and cross-domain settings. Furthermore, we design a molecular self-supervised pre-training framework aimed at improving generic transferability for molecular property prediction. The pre-trained model fully captures 2D topological and 3D geometric information of molecules, enabling fine-tuning for different downstream property prediction tasks. Finally, we design a mask prior-guided denoising diffusion framework that improves generic transferability for inverse protein folding, which involves iteratively generating feasible amino acid sequences that can fold into a given protein structure. In this thesis, extensive experiments are conducted to demonstrate the effectiveness of our proposed frameworks compared to related state-of-the-art methods. We also identify potential research directions in this emerging field for future exploration
Methods to understand the whole life carbon implications of school retrofit at scale
Get PDFThe built environment directly accounts for 25% of UK emissions, the majority of which are caused by energy consumption from existing buildings. Mass intervention is required, including replacement of existing fossil fuel heating systems and improvements to fabric efficiency.
In fact, 97% of stock within Europe needs upgrading to meet 2050 emissions targets.
As operational emissions are reduced, the embodied impact of building materials will account for a larger proportion of future emissions. As such, whole life carbon retrofit assessments are advocated to improve our understanding of how to meet carbon targets. A large proportion of current work focuses on residential properties and fabric retrofit measures. When applied at scale, studies often use an archetype approach with buildings grouped into specific age and form categories.
Non-residential buildings often have an increased geometric complexity. This work demonstrates a parametric approach, linking building form to retrofit decision making from a whole life carbon perspective. The results illustrate how building form influences different intervention decisions. The comparison of retrofit to demolition and new construction shows only the most inefficient building forms would benefit from replacement.
A study of English school stock demonstrates the impact of refurbishment on total emissions compared to a devised carbon budget. Results show pathways in which the carbon budget can be met, highlighting the impact of including building form in retrofit decision making.
This thesis shows that retrofit of schools must increase, 10 - 50 times from current practice, to meet the 2050 Climate Change Committee (CCC) carbon budget. The retrofit rate is reduced by 20% by prioritising the least thermally efficient buildings in both fabric and form. The high embodied carbon cost of refurbishment, especially heat pumps, prevents school stock meeting its most stringent, Tyndall centre derived, budget even if all buildings are retrofit in 2025
Implementation of a Competency Framework in Saudi Undergraduate Medical Education
No full textIn 2011, a competency-based medical education (CBME) framework called SaudiMEDs was developed in Saudi Arabia to serve the community with medical graduates who can competently fulfil society’s health needs. SaudiMEDs serves as a guide to designing, developing, and evaluating medical school curricula. Despite the widespread implementation of SaudiMEDs, no research has been conducted on how medical education has been shaped as a result of SaudiMEDs. This study’s primary aim is, therefore, to investigate the real-time implementation of SaudiMEDs and its impact on Saudi medical schools.
Qualitative case studies with two medical schools across the country were conducted to investigate how SaudiMEDs had been implemented. Documentary analysis of the medical schools’ documents, semi-structured interviews with academic leaders, and focus groups with faculty members and students were carried out for each school. Framework analysis for documents and Reflexive thematic analysis for interviews and focus groups were used to identify patterns and concepts and to conceptualise and construct meaning from the data. Analysis revealed a complex network of mixed perceptions of SaudiMEDs which, based on identified variations in teaching strategies, ways of learning and assessment methods, are considered as a key influence shaping implementation of SaudiMEDs. The perceptions of faculty members and students provide insight into the culture of each medical school and how it mediates implementation.
The study also identifies different challenges in the SaudiMEDs implementation, which could provide further explanation for the variation in each medical school context. A key finding was the challenge of translating curriculum innovation developed in and for Western, specifically North American, contexts to the distinct organisational and institutional culture of Saudi Arabia. One consequence of this was that schools in this study approached implementation as a mapping activity and focused on the short-term accreditation standard requirement rather than a longer-term evaluation. This limited the ability of SaudiMEDs to achieve its stated aims. The research helps to identify what needs to be done to enable the fruitful transformation of SaudiMEDs into robust curricula: a) a rigorous review of SaudiMEDs to contextualise community health needs and b) support for CBME implementation through the enactment of the legislation, training, and establishment of appropriate information technology infrastructure to provide authentic experiences for learners
Graph state properties and applications in the stabilizer formalism
Get PDFGraph states form a class of entangled quantum states that have multiple useful applications within quantum computing and quantum communication protocols. The stabilizer formalism offers an efficient mathematical description of graph states and the effect of operations acting on such states. Here, we consider three scenarios where the stabilizer formalism can be utilised to investigate the structure, manipulation and generation of graph states. First, we propose a method to calculate the purity of reduced states of graph states entirely within the stabilizer formalism, using only the stabilizer generators for a given state and apply this method to find the Concentratable Entanglement of graph states. Next, we reduce the number of qubits required within a graph state used as a resource for the measurement based implementation of a general two-qubit unitary, using the stabilizer formalism to track supplementary operations that are required. Finally, we examine spin-photon interactions followed by single-qubit measurements as a process to probabilistically generate states described by the stabilizer formalism, particularly when the phase shift induced is small
Pharmaceutical Pollution in Agricultural Environments: Fate, Behaviour, Uptake by Crops, and Implications for Human Health
No full textPharmaceutical pollution is ubiquitous in the environment, and there is widespread concern over the potential effects on human and ecological health. Agricultural systems are of particular concern due to their proximity to multiple sources of pharmaceutical contaminants and their position at the interface between human activities and natural ecosystems. Whilst numerous studies have examined pharmaceutical uptake by crops, inconsistent methodologies and limited understanding of influencing factors have hindered the ability to draw cohesive conclusions and effectively utilise current data.
This thesis investigates the key factors influencing pharmaceutical uptake by crops and the associated risks to human health. A novel database, the Uptake of Pharmaceuticals Into Crops (UTOPIC) database, was developed to systematically compile global evidence on pharmaceutical uptake by crops. The UTOPIC database was applied to a systematic review and meta-analysis to evaluate the current state of the science and elucidate the factors affecting pharmaceutical uptake by crops. The UTOPIC database also informed a human health risk assessment, evaluating potential exposure through consumption of contaminated crops.
Findings indicate that soil physicochemical properties are a key factor influencing pharmaceutical behaviour in the environment. Despite this, few studies accounted for soil variability in their experimental designs. Targeted experimental investigations demonstrated that pharmaceutical fate and crop uptake varied significantly with soil type, with soil porewater concentrations emerging as a key determinant of bioavailability and subsequent uptake.
These insights have critical implications for environmental risk assessments, which often do not account for the potential variability across diverse soil systems. Overall, this thesis advances understanding of pharmaceutical behaviour in agricultural environments and provides essential tools and recommendations for more accurate assessment and management of pharmaceutical pollution in the environment
Using phenomenology and semiology to support the differential diagnosis of transient loss of consciousness
No full textBACKGROUND: Transient loss of consciousness (TLOC) is a common acute presentation; over 90% are due to either syncope, epilepsy, or functional/dissociative seizures (FDS). Differential diagnosis is challenging. Better clinical criteria to support differential diagnosis – including clinical decision aids (CDAs) – could improve outcomes.
OBJECTIVES: This thesis aims to: (1) review barriers to accurate TLOC diagnosis; (2) provide external validation of candidate diagnostic criteria; and (3) develop a CDA for TLOC diagnosis.
METHODS: Methods include: (1) narrative review; (2) ethical analysis; (3) thematic analysis of semi-structured interviews with 20 first-presentation TLOC patients; (4) retrospective cohort video study of 189 videos from 50 patients with epilepsy or FDS; (5) retrospective cohort diagnostic accuracy study of 300 patients with syncope, epilepsy, or FDS; and (5) prospective cohort diagnostic accuracy study of 178 first TLOC patients.
RESULTS:
Experiences of initial TLOC assessment: First TLOC can be a disorienting ‘biographical disruption’. Communication supports interim self-management.
Individual diagnostic features: FDS are more likely than epilepsy to show peri-ictal social responsiveness in video review. This is externally validated as predicting FDS in first-presentation TLOC, as are fluctuating course or waxing/waning movements, asynchronous limb movements, younger age at onset, and total non-ictal and peri-ictal symptoms. Other diagnostic features derived from chronic patient cohorts are not validated in first presentations.
CDA development: A machine learning classifier trained on 36 patient/witness questionnaire responses in the chronic syncope/epilepsy/FDS cohort classified 86.0% of patients correctly. Validation accuracy was worse (75.8%) in the first-presentation TLOC cohort. A classifier trained on first-presentation data identified 9 optimal patient-reported predictors. It correctly identified 80.8% of diagnoses, non-significantly superior to initial clinician assessment (70.5%, p=0.192). Patients reported the CDA accessible and acceptable to use.
CONCLUSIONS: CDAs could improve outcomes for patients who experience TLOC. Future research needs to be conducted within the first presentation setting to ensure validity
Investigating the cellular landscape of glioblastoma brain tumours, and how it changes through treatment
Get PDFGlioblastoma (GBM) is the most aggressive primary malignancy of the central nervous system. Despite standard treatment, comprising surgical resection, followed by concomitant radiation and chemotherapy, it is incurable. This devastating prognosis stems from complex multi-layered heterogeneity, which enables GBM tumour cells to resist treatment and reoccur. Advances in genomic technologies have classified GBM tumours at the single-cell resolution, revealing that malignant GBM cells occupy distinct neoplastic cell states which resemble neuro-developmental hierarchies and wound healing programs. These states are supported by complex interactions which include immune, healthy brain and vasculature cells.
To investigate how the cellular landscape of GBM tumours changes through treatment, I utilised an extensive dataset of paired (pre- and post-treatment) GBM tumour patient samples. These samples were profiled using bulk RNA sequencing (RNA-seq) and thus characterising their cell type composition in silico, necessitated the use of cellular deconvolution techniques. Benchmarking of such methods has shown that accuracy and interpretability are highly dependent on the specificity of the cell type reference used. Therefore, I developed a set of GBM-specific cell type markers and used these to validate the optimal deconvolution method, which I also released as a publicly available web application, GBMDeconvoluteR. Using this tool, I characterised 219 paired GBM samples and uncovered consistent cell type changes through treatment. These changes were associated with survival outcomes and aligned with our previously described patient stratification, based on treatment-resistance mechanisms.
To complement these findings, I then applied a novel spatial proteomics method and found that hypoxia drives the layered organisation of the GBM tumour microenvironment (TME) pre-treatment, but post-treatment the GBM TME is less structured, driven instead by reactive astrocytes and infiltrating lymphocytes. Collectively, this work highlights some key shifts in the cellular landscape of GBM through treatment, which may hold therapeutic potential
Fibre fragmentation and release from textiles during laundering: Key influencing factors
No full textExoticism and Epistemic Resistance: Angela Carter’s (De)Colonial Journey
No full textAs a white female traveller and writer, Angela Carter occupies a complex positionality – both privileged and marginalised, simultaneously exoticised and exoticising during her foreign encounters. Hence, Carter’s self-proclaimed feminist radicalisation needs to be situated against this complex positionality, including the ways in which she effeminises her Japanese lover. This ambivalence finds wider expression in Carter’s depictions of non-Western Other(s), where she subverts but at times also reinforces a Eurocentric gaze that is coded as white and masculine. While previous scholarship has focused primarily on Carter’s Japanese experiences, this study self-consciously expands the scope to examine her cross-cultural engagements with Chinese philosophy, Indigeneity, and Black femininity in her construction of exoticism. By analysing her published works alongside private journals and correspondence archived at the British Library from feminist, postcolonial, and decolonial perspectives, it reassesses Carter’s writing of exoticism as a fragmented, non-linear and uneven (de)colonial journey. Thus, the project contributes to contemporary discussions on decoloniality, intersectionality, and exoticism as a fluid and reciprocal discourse. In doing so, it constitutes a decolonial act of reading – one that uproots static cultural and epistemological frameworks, engaging Carter’s work as a site of both critique and resistance.
The thesis is structured around Carter’s various (de)colonial journeys. Chapter 1 examines her literary traversal of the European past in Heroes and Villains (1969). Chapter 2 starts with an analysis of Carter’s intellectual commentaries on Japanese culture; then it focuses on her philosophical journey into ancient Chinese philosophy in her novel The Infernal Desire Machines of Doctor Hoffman (1972). Chapter 3 delves into Carter’s evolving portrayal of Indigeneity in Doctor Hoffman, ‘Master’ (collected in Fireworks (1974)), and ‘Our Lady of the Massacre’ (collected in Black Venus (1985)), whereas chapter 4 investigates her rewriting of Black femininity in The Passion of New Eve (1977) and ‘Black Venus’ (1980)
Data sources:
Access Repository Dashboard
Do you manage Open Research Online? Become a CORE Member to access insider analytics, issue reports and manage access to outputs from your repository in the CORE Repository Dashboard! 👇