42337 research outputs found
Sort by
Learning shape, structure, and semantics: self-supervised learning with 3D priors
No full textThe world exists in three dimensions, yet when 3D objects are projected onto a 2D image plane, vital spatial information is inevitably lost. Despite this limitation, humans possess a remarkable ability to infer 3D structure from 2D images, enabling us to navigate and interact seamlessly with our surroundings. In contrast, modern computer vision algorithms primarily interpret the world as a collection of 2D patterns (e.g. bag of 2D visual words), leading to several shortcomings: poor generalization to novel environments, difficulty in learning object categories from limited training samples, and vulnerability to adversarial attacks, where minor texture modifications can drastically degrade performance.
This thesis aims to reduce the gap between human and machine perception by improving the extraction of 3D object shape information from 2D images and leveraging 3D understanding to enhance high-level vision tasks such as semantic correspondence estimation. To do so, we take inspiration from developmental psychology which suggests that human vision is strongly driven by shape cues, particularly in early cognitive development. However, with the rise of deep learning, classical approaches that explicitly encode shape, such as pictorial structure models and deformable part-based models, have largely been abandoned in favor of end-to-end learning paradigms.
In this thesis, we first assess the capabilities of unsupervised computer vision models on semantic correspondence tasks using a novel evaluation protocol that jointly captures semantic and geometric understanding. Our findings reveal that current models fall short on this task, and we proposed a new method that improved the state-of-the-art performance at the time, demonstrating significant advancements over existing approaches.
Next, we introduce a method for extracting the 3D shape of articulated objects, such as animals, from single-view images without requiring manual supervision. Finally, we present a novel approach to integrate 3D priors into self-supervised learning frameworks, improving robustness for semantic tasks such as image recognition while maintaining accuracy. By emphasizing the role of 3D shape in visual learning, this work introduces new methods that enhance the robustness of machine perception, advancing it toward human-level competence
Ecological approaches to education and educational research: relationality, reciprocity, and resilience
Get PDFIn the face of unprecedented social and environmental crises, there have been calls
for approaches to education which are fundamentally relational, emphasising our
interconnectedness with each other and with the world (Blenkinsop & Kuchta, 2024).
However, western education systems have been critiqued for prioritising individual
success and values associated with marketisation, moves which run counter to
fostering reciprocity and a sense of community that are crucial for sustainability
(Biesta, 2020). Educational research has developed along similar lines, with projects
focusing on how to maximise performance and increase test scores, and
researchers placed at a remove from the field rather than in relation with participants
and data (Hultman & Lenz-Taguchi, 2010).
In this thesis, I argue that ecological approaches to conceptualising educational
spaces and educational research can reorient learning and teaching towards values
rooted in our connection as human beings and our relationships with the world. I
critically engage with the theory of learning ecologies, which stems from the concept
of ecosystems in the biological sciences (Barnett & Jackson, 2020), and brings focus
to relationships between learners, their environment, and the different elements that
impact upon and are impacted by their learning (Jackson, 2013). Understanding
different educational spaces as an ecology can put our relationships into the
spotlight, asking questions of the quality of interactions for students, teachers, and
the wider environment. However, ecological models have also been described as
overly complex, with a need for further empirical evidence and clearer frameworks to
understand the concept in practice (Sangrà et al., 2019). These debates, dilemmas,
and opportunities call for further investigation of learning ecologies across varied
educational contexts: specifically indoor, outdoor, and online spaces.
I engage directly with these tensions by presenting a model for ecological
approaches to education and educational research to critically review seven of my
research project publications. The studies focus on a variety of educational contexts,
actors, and environments, addressing key UK and Scottish policy drivers (Scottish
Attainment Challenge; Learning for Sustainability; Scottish Government 1+2
Language Strategy; National Curriculum Guidance for EAL (English as an Additional
Language)) and their interpretation in practice. The studies are also outward-looking, as they address global issues of migration and environmental and educational
sustainability.
In the Introduction to this thesis (Chapter 1), I outline my conceptual model for
ecological approaches to education and educational research. Employing this
theoretical lens, I undertake a critical review of my three lead author publications
(Chapter 2) and four co-author publications (Chapter 3), highlight my contribution to
the research and writing, and consider impact and limitations. In the critical review I
apply reflexive qualitative thematic analysis (Braun & Clarke, 2019) using my
conceptual model, focusing on spatial and temporal dimensions, and concepts of
agency, identity, and uncertainty. My aim has been to draw light on the nature of
learning ecologies, how they can be defined and developed in a range of educational
contexts, and how they can be further researched and realised.
A critical review of my publications reveals that, despite their perceived complexity,
ecological approaches and considerations of space in learning and teaching can
surface the relational and reciprocal in education (Sepie, 2017), and capacities for
building resilience. An ecological framing situates educational research as an
ongoing dialogue and in a state of emergence and becoming, encouraging
engagement with uncertainty and the quality of relationships as they are unfolding.
Across my publications and the different educational spaces (indoors, outdoors,
online), an ecological lens reveals a plurality of ways of being and doing, reimagining
education as it could be.
I conclude by discussing the impact of my publications in relation to research on
educational spaces, learning for sustainability, and the influence on policy and
practice in education. By emphasising values of relationality, reciprocity, and
resilience, I argue that developing educational spaces as healthy ecologies can bring
hope and possibility in the face of uncertain times
Oh, the humanity! A human-centric approach to social bias research in natural language processing
No full textMuch current research into social bias in Natural Language Processing (NLP) -- that is, the tendency for NLP technologies to reflect human biases such as sexism and homophobia in the relative probability of different outputs -- suffers from relying on a superficial understanding of the problem. The issue of social bias is treated as a mathematical kink that needs to be ironed out, after which the harm that the model does will be irrefutably reduced -- a form of algorithmic idealism. Social bias is seen as an unfortunate result of "dirty data'' and "data imbalance'', and practitioners typically focus on addressing social bias through changes at training or inference to counteract these data issues. Little regard is given to the human aspect: to the myriad normative choices made by those who develop these systems; the beliefs of those who deploy them; nor to the response of those impacted by these technologies, all of which will influence how social bias is actually experienced. Operationalising bias as a quantifiable metric allows for at-scale evaluation that keeps pace with the rapid development of new NLP technologies. However, I argue this superficial understanding of social bias will lead us to superficial and ultimately ineffective solutions, which ignore the role of human behaviour in determining the harm done by technology. As I demonstrate, heuristic attempts at social bias mitigation often end up doing more harm than good.
In my thesis, I advocate for a human-centric approach to measuring and mitigating social bias in NLP, one which focuses on human choices, human identities and human behaviour, to give a more complete understanding of the true impact of NLP technologies. A human-centric approach treats social bias as a socio-technical problem, and casts its net widely over a broad range of stakeholders, sources of bias, and demographic attributes. My proposed approach is underpinned by five maxims: see technology as part of a socio-technical system; consider many sources of bias; focus on the impact on people and how they respond; be driven by social science theory and community knowledge; address a broad range of demographics.
I present my work as four case studies across three tasks which demonstrate the benefits of this approach. I consider harms against marginalised (primarily queer) identities through social bias in sentiment analysis tools, text-to-image (TTI) models and social media recommender and moderation algorithms (namely on TikTok), finding that heuristic attempts to reduce social bias often do more harm than good, and that the public form complex beliefs about NLP technologies. In all my work, I address social biases in publicly available or public facing tools, as these typically have a broader impact than state of the art models. I focus primarily on harms done to the LGBTQ+ community, in part because of personal relevance, but also because it provides an opportunity to demonstrate the benefits of an approach that considers demographic qualities beyond a binary. There are no binaries in nature -- in human identity -- yet much social bias research hinges on treating demographics as such. In doing so I contribute significantly to our understanding of queerphobia in NLP.
Ultimately I argue -- despite the title of this thesis -- that the best approach to social bias research is one that switches focus from social bias to real-world harms. Social bias has been used as a proxy for harm, but as I argue, it is often a very poor one. Changing our focus to harms necessitates defining specific use contexts (real or imagined). The "meaning'' of a difference in probability will be context dependent, as will how this difference is interpreted by those impacted by the model. I am far from the first to critique current practices, and the disconnect between social bias and harm; I amplify the message, and enrich it with five clear maxims that improve the validity of social bias research. To leave the human context out of the equation when measuring harm is nonsensical, yet for too long the field of NLP has attempted to do exactly that. Oh, the humanity
The story of the three bloods: the effect of coagulation on drying blood droplets
Get PDFIn order to pursue criminals as quickly and efficiently as possible, forensic analysts
must glean a wealth of information from a limited set of evidence in a short
amount of time. This research aims to expand the information acquired from
a crime scene by increasing understanding of blood drying behavior. Blood
is a mixture of biological colloids and proteins, the drying of which has been
examined for some time. These experiments overwhelmingly use anticoagulants
such as EDTA and tri-sodium citrate to extend the lifetime and transportability
of the blood by preventing coagulation, making it easier to perform laboratory
experiments. While experiments using anticoagulated blood produce useful
information for personalised medicine, they may be of limited relevance for the
typical crime scene. To extend the usefulness of blood droplet research to reallife
crime scenes where coagulation naturally occurs, fresh blood coagulation and
a coagulation-like process utilised by the medical community for clotting assays
were introduced to small whole blood droplets. Gravimetrics, optical coherence
tomography, and image analysis of drying time-lapses for various geometries were
used to examine three major areas: evaporation dynamics, cracking patterns, and
substrate adhesion. Experiments found important differences in the delamination,
internal dynamics, and final morphology of the droplets which may impact
forensic conclusions, though the drying time and evaporation rate were found
to be identical across all treatments
Raman microscopy and isolation of aged yeast cells to investigate cellular lipid changes in aging
Get PDFLife expectancy increased to about 80 years, but advanced age is still a risk to human beings. To deal with this risk, many studies worked on the mechanisms behind aging and senescence and relevant diseases. Aging refers to organism-wide functional decline, senescence occurs at the level of individual cells entering permanent cell-cycle arrest. Accumulations of senescent cells are linked to metabolic imbalances, neurodegenerative diseases, and other age-associated disorders, with the dysregulation of lipid metabolism playing a particularly important role. Lipids serve structural functions in cellular membranes and act as key mediators in signaling, transport, and metabolic pathways. During aging, these lipids can accumulate oxidative damage and peroxidation by-products, contributing to chronic inflammation and organ dysfunction. Understanding lipids behavior might contribute to the knowledge of aging progress.
This dissertation focuses on two main objectives. First, utilize Raman Microscopy to mapping and monitor intracellular lipids as it is examined as a powerful, non-invasive tool to characterize molecules. Second, budding yeast (Saccharomyces cerevisiae) is employed as a tractable model to investigate mechanisms of cellular aging, particularly the isolation of older cells that display key senescence markers. By combining Raman spectrometric analysis with refined yeast isolation strategies, this work aims to deepen our understanding of how lipid disruptions intersect with aging. The hypotheses guiding these studies are that (1) Raman-based mapping can reveal dynamic lipid changes within individual cells over time, and (2) it is feasible to isolate old yeast cells of different replicative ages to investigate the lipidome’s role in senescence. However, the results show that there is a limit for lipid real-time mapping, and the yield of isolated old budding yeast cells by magnetic sorting is low, and some optimization might increase the yield. Ultimately, the results and discussion offer new perspectives in optimizing lipid mapping by Ramam Microscopy and the isolation of aged budding yeast cells
Data science for health technology appraisal: data intelligence for the better evaluation of new treatments for breast cancer
Get PDFBACKGROUND:
Breast cancer remains the most common malignancy and leading cause of cancer-related death among women worldwide. While localised disease is largely curable, metastatic or recurrent cases carry a poor prognosis and present significant challenges for healthcare systems. The NHS faces unprecedented pressures due to resource scarcity, capacity constraints, and the rapid introduction of innovative therapies, particularly in advanced breast cancer (ABC). Existing budget impact (BI) assessments, such as those provided by the Scottish Medicines Consortium (SMC), often rely on assumptions and expert opinion, failing to capture the complexity of real-world decision-making. This can result in either over- or underestimation of budgetary needs, leading to opportunity costs or unexpected debt. There is a pressing need for actionable real-world evidence (RWE) and more rigorous, data-driven methodologies to inform budget impact analysis (BIA) and support sustainable, value-based healthcare.
AIMS & OBJECTIVES:
The overarching aim of this research was to advance health data intelligence capability within NHS Lothian by developing a model-based, real-world data (RWD)-driven approach to BIA. This approach seeks to provide more accurate predictions of the cost implications and resource demands associated with adopting new therapies for advanced breast cancer, thereby informing regional guidance and supporting feasible, affordable care provision. Specifically, this work leverages Scotland’s rich RWD assets and patient-level simulation modelling to deliver a structured, accurate BI forecasting tool for new drugs in Secondary Care. The objective was to develop a proof-of-concept budget impact model (BIM) using exemplar new drugs for ABC within the Edinburgh Cancer Centre, providing a robust, data-driven solution to NHS budget pressures.
RESEARCH QUESTIONS:
1. What are the current methods for BIA in healthcare?)
2. How can we improve current BIA by better use of RWD and the application of new or improved methods?
3. Can new approaches improve the adoption process for an exemplar new treatment of advanced breast cancer in Scotland?
METHODS:
The study targeted female post-menopausal patients diagnosed with ER+ / Her2- metastatic or locally advanced breast cancer between 2013 and 2017 within NHS Lothian, who were previously untreated for their advanced disease. Regional and national routinely collected data were linked to reconstruct and segment clinical pathways for CDK4/6 inhibitor-eligible ABC patients (N=74). Analyses were conducted within the National Safe Haven/ DDI DataLoch trusted research environment (TRE), extracting non-disclosive summary statistics and parametric model distributions. The BIM prototype, constructed and implemented in R/Studio, uses a probabilistic patient-level state-transition model as its structural core, integrating both published parameters and those derived from descriptive analysis of real-world patient-level data. The model captures incidence, prevalence, and cost dynamics over a 5-year time horizon, allows scenario analyses, and generates QALYs alongside financial estimates—thereby interfacing cost-effectiveness and BI modelling.
FINDINGS:
A user-friendly prototype BIM was developed, revealing substantial discrepancies between RWD-based and SMC BI template estimates. The SMC calculator significantly underestimated the 5-year BI (£5,851,911 for 1,700+ patients), compared to the BIM’s estimate (£4,045,092.27 for 121 simulated patients). Accurate BI estimation is crucial: overestimation risks opportunity costs, while underestimation can lead to overspending and debt. The new BIM successfully integrated QALYs and offers a flexible, scenario-analysis-based framework for more realistic BI forecasting for CDK4/6i-eligible patients. Findings also highlighted interoperability and analytical challenges, necessitating further validity checks and testing the generalisability of the BIM prototype.
CONCLUSIONS:
BIA provides policymakers and NHS managers with essential estimates of the financial and service implications of adopting new healthcare technologies, complementing cost-effectiveness analysis by offering a temporal and quantitative perspective on resource and cost changes. This pioneering study demonstrates how real-world data can be converted into actionable, evidence-backed insights that enhance existing economic evaluations. The proposed BIM addresses persistent unmet needs of Scottish health technology assessment by bridging the gap between value and affordability, supporting the NHS objective of delivering value-based healthcare while maintaining fiscal sustainability
A Tool-Integrated RAG Framework for Scottish Gazetteer Retrieval and Geographic Visualization
No full textThe large language models are developing fast these years. With its powerful natural language generation capabilities, they are helpful in summarizing and refining information. Gazetteer for Scotland is a comprehensive database of Scottish geography which has tens of thousands of entries with long and shorts texts in it. An architecture of connecting LLM and GfS can be proposed to take advantage of LLM to make information in GfS more usable and accessible. A retrieval augmented generation system is designed for this task, searching and generating ground response according to users’ input and returning a map using GeoJSON and Folium framework in GIS processing unit. With replaceable LLM api and embedding model slot, different models can be deployed to this system, which is flexible. The final performance is better with ChatGPT-4 and AllMiniLM-L6-V2 loaded. In the GIS processing unit, Deepseek has the highest quality of response and ChatGPT has the fastest response time. The pipeline is considered efficient and accurate for most cases. However, there is still room for improvement in terms of transferability and long-text processing. This pipeline will be a basic prototype possessing architecture of dual output (natural language and GeoJSON) RAG agent system for future studies
Co-performative aesthetics of care: a posthumanist approach to trauma-sensitive generative meaning-making with care experienced communities and lens-based technologies
Get PDFThe Aesthetics of Care explores the conditions that shape trauma-sensitive, collaborative meaning-making with care experienced communities. Care experienced refers to individuals who are currently in care or have lived in care. These communities often face reductive narratives shaped by deficit-based data and services designed for them rather than with them. Taking a queer and critical stance, this research challenges such framings by exploring generative (co-creating new insights through making and imagining) approaches that centre lived experience and value relational, imaginative, and performative ways of engaging with people, technologies, and environments. My background in design and participatory photography with communities who experience stigma informs this approach. In particular, the analogue photographic process, where images remain invisible until developed, serves as a conceptual anchor for the aesthetics of latency. This metaphor frames a research process that values uncertainty, emergence, and the unfolding of meaning through relationship and time.
The theoretical foundation of the study draws on posthumanist thinking, particularly Karen Barad’s agential realism, which views people, technologies, environments, and ideas as entangled and mutually co-constitutive. From this perspective, meaning is not pre-existing or extracted but emerges through co-performative relations that hold space for complexity, care, and the potential to become otherwise. This worldview underpins the methodological framework for this PhD, Research through Design (RtD), extended as a trauma-sensitive mode of inquiry, where knowledge is generated through cycles of making, reflecting, and imagining in response to values and themes identified by care experienced participants. It prioritises emotional safety, emancipatory practice, and care, supporting ethical and situated modes of collaborative inquiry that make space for the emergence of new meanings and the imaginative expression of possible futures.
The research design extends Liz Sanders’ generative design methods, developed initially within product and service design, to focus on the relational qualitative process of meaning-making. This approach is informed by the shared anthropology of ethnographic filmmaker Jean Rouch, which values playful improvisation and fiction as means of co-creating meaning and generating participant-defined values and dreams. Generative design focuses on illuminating tacit knowledge, which is embodied and experiential, and latent knowledge, which often remains unspoken or unseen until developed through speculative and creative practice. Through a series of RtD workshops in Edinburgh and London, care experienced young people co-created speculative scenarios with fictional characters, improvisation, and interactive lens-based technologies, including three-dimensional imaging and augmented reality. These encounters generated prototype digital epistemic artefacts, creative outputs representing the expression, development, and communication of situated knowledge. The application of verbatim recordings and fictional distancing reduced the risk of re-traumatisation and created space for emotionally rich stories to emerge as speculative possibilities within a situated, relational practice.
The findings illuminate how material-discursive factors, understood as the inseparable entanglement of material conditions, technologies, environments, language, and values, shape the process and practice of generative meaning-making. In presenting a new trajectory for co-creating and communicating data and values with communities that are stigmatised and with policymakers, the research makes the following contributions. First, this study advances generative design research and Research through Design (RtD) by integrating trauma-sensitive and posthumanist considerations, offering conceptual and practical insights for collaborating in ethical inquiry with care experienced communities, particularly in illuminating tacit and latent domains of knowledge. Second, it contributes to human-computer interaction (HCI), particularly Fourth Wave scholarship, by demonstrating how lens-based technologies can function as relational and co-performative constituents in participatory and speculative meaning-making. Third, it extends contemporary visual ethnography, building on Sarah Pink’s emphasis on sensory and embodied approaches, by showing how co-performative and multisensory artefacts can communicate complexity and engage stakeholder audiences, including policymakers, in inclusive and affective ways. By proposing embodied listening and playful co-performativity with digital media as a trauma-sensitive mode of care, this thesis presents an ‘Aesthetics of Care’ approach, a generative model for collaborative research that honours the complexity of lived experience and invites new relational understandings to emerge by offering transferable methodological foundations for trauma-sensitive design practice.
The practice-based outputs of this research can be found at ethnofiction.xy
Endothelial cell regulation and function in development and disease
No full textCardiovascular diseases (CVDs) remain a leading global cause of mortality, partly due to the limited regenerative capacity of cardiovascular tissues. The vascular endothelium, a monolayer of specialised cells, is essential for maintaining vascular tone, barrier function, and immune regulation. Dysregulation of endothelial cell (EC) function is a central contributor to CVD progression. This thesis investigates EC regulation in two distinct contexts: first, the potential transcriptional maintenance of arterial identity during development, and second, the response of the endothelium of the pulmonary microvasculature to severe COVID-19.
While several pathways involved in arterialisation are known, the transcriptional mechanisms that may maintain arterial identity remain incompletely defined. Using single-cell RNA sequencing data from human stem cell-derived ECs, fetal heart tissue, and annotated public datasets, eight transcription factors (TF) or proteins associated with transcriptional pathways were found to be enriched in arterial EC populations: BCL6B, MAFF, PRDM16, SOX7, SMAD6, HES4, TCF4 and EPAS1.
The second part of the thesis presents published work examining pulmonary microvascular endothelial cell responses to plasma from patients with severe COVID-19. Treated cells exhibited reduced viability, impaired barrier function, and widespread subcellular remodelling. Transcriptomic profiling revealed consistent dysregulation of transcriptional regulators and components of TGF-β signalling, suggesting that exposure to inflammatory plasma perturbs key endothelial regulatory networks. These findings provide insight into how severe COVID-19 may contribute to long-term vascular complications.
Together, this work highlights the relevance of transcriptional regulation in both the maintenance of endothelial identity and its disruption during severe disease. By examining these processes in parallel, it offers groundwork for understanding of how EC function may be altered and potentially supported across developmental and pathological settings
Dissecting the genomic and epigenomic alterations underlying glioblastoma
No full textGlioblastoma (GBM) is the most common intrinsic primary brain tumour, characterised by structural complexity, tumoural heterogeneity and poor prognosis. GBM tumours
reactivate neurodevelopmental programmes and they are driven by GBM stem-like cells
(GSCs), displaying phenotypic similarities to neural stem cells (NSCs). Although much is
known about the recurrent coding alterations and the role of GSCs in GBM, we still have
a poor understanding of the contribution of non-coding alterations and complex structural variants (SVs) to GBM pathogenesis, as well as the role of key NSC transcription factors such as SOX2 and SOX9, in regulating self-renewal activity in GSCs and gliomagenesis.
To study non-coding alterations and complex SVs, we compiled a large (n = 230) cohort
of GBM tumours with whole genome sequencing data. We uncovered a lack of recurrent
non-coding somatic small mutations (SSMs) apart from TERT promoter mutations. We
identified loci under putative selection for SV recurrence, including those harbouring focal amplications and deletions, associated with complex SVs such as extrachromosomal DNA (ecDNA) and chromothripsis. We discovered underappreciated features of ecDNAs in GBM, including a high prevalence of ecDNAs generated via episomal exclusion, and uncovered the extent by which ecDNAs shape tumoural heterogeneity via seismic amplification and putative chromosomal re-integration. Notably, despite the relatively high SSM- and SV-derived neoantigen burdens, ecDNA+ samples exhibited higher transcriptional immune suppression compared to ecDNA- samples, an effect not seen in other complex SV types, suggesting that ecDNAs may modulate the tumour-immune microenvironment to evade immune clearance. Last, contrary to our expectations, ecDNA and chromothripsis were not associated with poor overall survival; rather, patients whose GBMs sustained high levels of overall complex SV burden were associated with significantly shorter overall survival, independent of other available prognostic factors.
To complement these genomic analyses, we also explored transcriptional regulatory programs in GBM, particularly the role of SOX2 and SOX9 in regulating GSC self-renewal. We used chromatin immunoprecipitation followed by sequencing (ChIP-seq) data across 7 patient-derived GSCs. We found that SOX9 binds exclusively and at close proximity to SOX2, and co-bound SOX2/SOX9 sites are enriched with GSC-specific super-enhancers (SEs). We identified a dual-mode of SOX2/SOX9 co-binding that was regulatory element-dependent – as a monomer at promoters and a dimer at enhancers, reminiscent of SOX9 activity in cartilage development. Co-bound SOX2/SOX9 SEs target genes regulating NSC identity, including QKI, PTPRZ1 and CDK6. Intriguingly, SOX2/SOX9 co-binding also target their own and each other’s regulatory regions, suggesting the existence of an auto- and cross-regulatory feed-forward axis. Importantly, we observed no increased in mutational burden across SOX2/SOX9 co-binding sites, arguing against mutation-driven reactivation processes at these neurodevelopmental enhancers. These findings altogether suggest that SOX2 and SOX9 cooperate at shared enhancers and target genes, and their increased co-activity may contribute to the core programs of self-renewal displayed by GSCs.
In summary, we present a thorough computational genomic analysis of the non-coding
alterations and structural complexity shaping the GBM genome, revealing a complex
interplay between ecDNA, immune evasion pathways and complex SV burden during
gliomagenesis. Furthermore, we identify putative downstream targets of SOX2 and SOX9 that may underlie self-renewal activity in GSCs. It is hoped that these findings may lead to an increased understanding of GBM pathogenesis and ultimately result in new
therapeutic avenues for GBM patients
Access Repository Dashboard
Do you manage Edinburgh Research Archive? Access insider analytics, issue reports and manage access to outputs from your repository in the CORE Repository Dashboard!