42337 research outputs found
Sort by
Evaluating the Proxy Potential of Land Use and Macroinvertebrates for Water Quality Assessment in the Lower Mekong Basin
No full textEffective water quality monitoring in the Lower Mekong Basin (LMB) is challenged by the region’s vast spatial extent, limited station coverage, and the logistical complexity of chemical sampling. Yet, given the basin’s ecological significance and the millions of people who depend on its freshwater systems, scalable approaches to aquatic assessment are critically important. This study evaluates whether upstream land use patterns and downstream macroinvertebrate communities can reliably serve as proxy indicators for surface water quality in the LMB.
Using dry-season data from 34 monitoring stations spanning tributary and mainstem sites, the study integrated geospatial land cover metrics, physicochemical water quality variables, and benthic and littoral macroinvertebrate metrics to explore proxy relationships. Generalized Additive Models (GAMs), Redundancy Analysis (RDA), and variance partitioning were employed to assess predictive performance across parameters including total suspended solids, pH, dissolved oxygen, nutrient concentrations, and chemical oxygen demand.
Results indicate that both land use, particularly urban and agricultural cover, and macroinvertebrate metrics, such as ATSPT and richness, are independently and complementarily associated with water quality variation. Integrated models (with land use and macroinvertebrates as predictors) showed the strongest explanatory power, with proxy strength varying by water quality parameter. Temporal analysis revealed changing macroinvertebrate composition and increasing anthropogenic land use, underscoring the importance of multi-year monitoring.
By demonstrating the viability of spatial and biological proxies, this study supports a more scalable framework for freshwater assessment. It offers basin-scale insight into ecological condition and lays the groundwork for proxy-based monitoring strategies that are accessible, cost-effective, and applicable to other data-limited river systems
Geospatial UK: Linking Scotland to a Geospatial Future
No full textThe geospatial industry is one of the fastest growing industries in the United Kingdom, and Scotland in particular is looking to leverage this boom in order to become a major geospatial hub within Europe. However, the majority of businesses in this sector report issues with both hiring and retaining adequate numbers of geospatial professionals to complete their work, creating a skills gap or shortage. Effort must be made to not only increase the number of youth who are choosing careers in the geospatial industry, but also to diversify the interests and careers of geospatial professionals. The best way to accomplish this is to incorporate geospatial skills, concepts, and data throughout Scottish secondary education, as well as market geospatial careers directly to students. The Geospatial UK website has been identified as a potential resource for students, teachers of all subjects, and careers counsellors to gain a deeper understanding of the role that geospatial plays in our lives, and the different pathways for entering the industry. Four market research style surveys were sent out to the following groups: students, Geography teachers, non-geography teachers, and careers counsellors. Responses to the surveys, as well as a review of relevant literature, informed the changes and alterations made to the site, including adding further academic resources and showcasing the diverse range of geospatial careers. Based on the survey results, additional outreach is recommended for non-geography teachers and careers counsellors to overcome knowledge gaps, and further updates to the site are proposed
Retrofit Option Scoring Matrix
Full text linkA plain-English, self‑guided scoring worksheet designed to help owners of historic and traditional properties compare retrofit options based on their priorities. Seven criteria are listed with brief prompts for consideration. Users can also write in their own criteria. There are rooms for scoring as well as notes, providing owners an organised space to compare the trade-offs of different options before making a decision. The worksheet is tailored for window improvements but can be adapted for other retrofits. When used to prepare for a window improvement, the worksheet is best read with “Retrofitting Historic Sash and Case Window – A Quick Start Guide for Homeowners” also available in ERA (https://era.ed.ac.uk/handle/1842/41922)
The genetic basis of previously unexplained aniridia
Full text linkClassical aniridia [MIM 106210] is a rare, panocular malformation characterised by a spectrum of iris and foveal hypoplasia, later complicated by cataracts, lens subluxation, glaucoma and keratopathy. It is overwhelmingly associated with PAX6 [MIM 607108] haploinsufficiency. PAX6 is a highly-conserved transcription factor that orchestrates eye development. Whilst the molecular genetic basis of aniridia and PAX6 disruption is broadly known, there remain unexplained areas that have guided the work in this thesis.
PAX6 missense variants can partially or fully phenocopy haploinsufficiency, causing classical aniridia. Why they also cause rare “worse than null” phenotypes is unclear. In order to study the effect of variant class on phenotype, I performed systematic, detailed ocular phenotyping followed by cluster analysis of PAX6 eye malformations. This compared all available missense variant cases (n=161) with a size-matched control of likely gene-disruptive (LGD) cases. Whilst LGD variants were almost universally associated with classical aniridia, the phenotypic spectrum of missense cases was wider and appeared to cluster separately, ranging in severity from isolated foveal hypoplasia (hypomorphic) to bilateral severe microphthalmia (worse than null). Recurrent substitutions of key residues in the DNA-binding paired domain of PAX6 accounted for the most severe cases, notably p.Ser54Arg and p.Asn124Lys. The impact of these variants was investigated in vitro, studying the binding of purified PAX6 paired domain to well-characterised PAX6 DNA targets, the LE9 and SIMO enhancers. Electromobility shift assays showed that binding was altered by p.Ser54Arg and p.Asn124Lys compared to wild type, though (unlike classical aniridia variants) not totally abrogated. Their impact differed between the two DNA targets studied. Whilst not explaining all of the observed genotype-phenotype correlation, these data suggest that the variants do alter protein-DNA interactions - consistent with results of protein modelling - and could affect the stoichiometry of PAX6 target gene activation.
I noted a striking similarity between the rare, severe PAX6 missense cases above and those seen in six families with heterozygous variants in MAB21L1 [MIM 601280], all resulting in Arg51 and Phe52 substitutions. All had aniridia and/or microphthalmia, 3/6 had both, and 2/6 had anomalous optic discs. MAB21L1 is involved in eye development downstream of PAX6, but was not known to cause aniridia. To characterise this further, ocular phenotyping of a CRIPSR-Cas9 gene-edited mouse strain carrying Mab21l1 p.Arg51Leu was performed. Adult mice heterozygous for this allele had a highly-penetrant excavated optic nerve anomaly. The phenotype overlaps with but is milder than the human counterpart. The homozygous animals showed severe microphthalmia. There are still questions to be answered regarding the mechanisms in both human and mouse, which may differ, but the human genetic data strongly support this being a new aniridia-associated gene with a gain of function effect.
A small proportion of aniridia is still genetically uncharacterised, indicating that there could be novel ways of inactivating PAX6 or additional disease loci. To address this, I undertook analysis of 37 unrelated mutation-negative individuals with classical aniridia (n=51 including relatives) using paired-end, short-read whole genome sequencing (WGS) technology. Following an in-house analysis pipeline, causative variants were identified in 22/37 cases (60%). 15/22 were structural variants, including: chromosome 11 inversions, one separating PAX6 from its critical downstream enhancers; deletions of this downstream enhancer region; whole/partial gene deletions of PAX6, and lastly three FOXC1 gene deletions. The remainder were intragenic PAX6 single nucleotide variants, including two deep intronic splice variants. RT-PCR analysis of RNA from patient-derived lymphoblastoid cell lines was performed in a subset, confirming missplicing in vitro. Notwithstanding historical differences in PAX6 screening, it is remarkable that over half the cohort had a disease mechanism involving either PAX6 or its regulatory region, highlighting the unusually strong disease-gene link.
In summary, this thesis fills gaps in our understanding of aniridia genetics through whole genome diagnosis of unsolved cases, underlining the importance of the wider PAX6 locus and adding to the known variants which inactivate it, as well as through characterisation of severe aniridia-microphthalmia phenotypes caused by rare missense variants in PAX6 and in the novel aniridia-associated gene MAB21L1
The epidemiology of bovine viral diarrhoea virus (BVDV) in Tanzanian smallholder dairy cattle
No full textSmallholder dairy cattle are an important source of livelihood for many rural households in
Tanzania providing power, food and income. The cattle and their farmers experience
numerous challenges from infectious diseases such as bovine viral diarrhoea (BVD). BVD
negatively impacts the livelihoods of the rural communities which make up about 70% of the
Tanzanian population, affecting food security and economic stability. Understanding the
epidemiology of BVD and its consequences is critical for both human and animal health,
thereby making it a crucial public health concern.
BVD caused by bovine viral diarrhoea virus (BVDV) is one of the most important diseases of
cattle. It has a global relevance and negative impact on production, reproduction and overall
herd health. It establishes an in-utero infection during the first trimester of gestation, resulting
in the birth of a persistently infected (PI) animal. PI animals act as superspreaders in the herd,
continuously shedding the virus via virtually all body secretions and excretions, including nasal
discharges, saliva, semen, urine, tears, milk, and skin and ear notch. The challenges caused by
BVDV include clinical signs such as gastrointestinal disorder, respiratory illnesses, foetal
defects, stillbirths, abortion, decreased milk yields, and mucosal disease (MD). The virus
causes immunosuppression, making infected animals more susceptible to other infections and
allowing their effects to be more severe.
Risk factors responsible for the transmission of the virus include larger herd size, animal age,
introduction of new animals, and poor biosecurity on the farm. Diagnosis of BVDV can be done
using a variety of antigen and antibody detection tests to help identify PI animals or transiently
infected (TI) animals.
Despite the high prevalence of BVDV in Tanzania, there is currently no national control or
surveillance for the virus. Control is possible using interventions such as vaccinations, testand-
cull, quarantine, and proper biosecurity measures on farms. A combination of these is
more effective than any single intervention on its own in controlling BVDV.
Modelling as a control tool can be done in the absence of interventions in a region as it allows
researchers to simulate different control scenarios, estimate infection spread and investigate
the best control strategy for a specific region.
Economic consequences such as low birth rate, high death rate of PIs, low milk production
and high costs of treatment are often associated with the virus. BVDV is poorly quantified in
many countries but there is a complete lack of evidence about the economic and financial
losses of the disease in Tanzania.
This thesis highlights and addresses some of the knowledge gaps in the understanding of the
epidemiology of BVDV in low-and middle-income countries (LMIC). It also investigates the
epidemiology of BVDV in Tanzanian smallholder dairy farms, with a focus on seroprevalence,
significant hotspots, risk factors, and control strategies relevant to the system in six key dairyproducing
Regions of Tanzania. Several analytical methods are used to study the epidemiology
of BVDV without control, the seroprevalence estimates, the most important risk factors
driving the virus spread, models to study the transmission of the virus and the most feasible
control methods to tackle the virus. Together, the results from the various chapters provide
methods and recommendations for evidence-based policies for the control of the virus.
We study the epidemiology of BVDV in Tanzanian smallholder dairy with the specific study
objectives outlined below:
Objective 1 - to synthesize existing knowledge on the prevalence, health impact, risk factors,
and economic consequences of BVDV in low- and middle-income countries through a
systematic review.
Objective 2 - to estimate the seroprevalence and investigate the spatial distribution of BVDV
in six key Regions of Tanzania.
Objective 3 - to identify risk factors associated with BVDV seropositivity and estimate their
strength of association with the outcome in smallholder dairy farms.
Objective 4 - to model the potential impact of vaccination on susceptible animals and culling
of PI animals on BVDV transmission within Tanzanian dairy herds.
In the first analysis, Chapter 2, a systematic literature review and meta-analysis was conducted
to identify current research gaps in literature in low- and middle-income countries (LMIC) for
BVDV prevalence, risk factors, health and economic impacts. The study identified gaps in
literature, particularly the scarcity of studies on the health and economic impact of BVDV in
these settings. While there are few studies estimating the seroprevalence of BVDV in LMIC
there are no studies for identifying BVDV risk factors driving the infection, and no models for
controlling the virus in Tanzania.
In Chapter 3, using the dataset available from the Centre for Tropical Livestock Genetics and
Health (CTLGH), I conducted an antibody enzyme-linked immunosorbent assay (ELISA) test to
obtain seroprevalence results and investigated the spatial pattern of BVDV in the six Regions
studied. The results revealed a BVDV seroprevalence of 25.0% (95%CI:23.1-26.9), indicative of
widespread exposure in smallholder dairy herds in the various Regions and identified the
areas where the infection was significantly clustering. These results showed that higher
seroprevalence was found in the Northern Regions and lower levels of seroprevalence were
found in the Southern Regions. This suggests more virus circulation in the north compared to
the south, and therefore targeting the northern areas first might be appropriate in terms of
control if resources are limited.
In Chapter 4, I identify the risk factors associated with BVDV seropositivity in Tanzanian
smallholder dairy farms, using mixed-effect multivariable logistic regression (MMLR) models.
The overall model identified herd size, management practices, animal age and prior
vaccination for other diseases (foot-and-mouth disease (FMD), and contagious bovine
pleuropneumonia (CBPP)) as significant risk factors for BVDV seropositivity in an overall
model. The unusual association between BVDV and vaccination seemed interesting, and this
led to the creation of Regional models to re-examine the seroprevalence by Region. Regional
models identified possible endemic transmission in Kilimanjaro, and different drivers in the
other Regions, which led to the conclusion that there was vaccine-related seropositivity in
Arusha, Tanga and Iringa, while Kilimanjaro had more of an endemic transmission.
The overall results show that BVDV is endemic, and the risk factors are mainly driven by
animal age in Kilimanjaro, but by vaccinations in three of the other Regions. Among the
possible explanation for this were: 1) vaccines contaminated with extraneous agents, or 2)
poor biosecurity measures where there is unhygienic use of hypodermic needles between
animals during vaccination.
Finally, in Chapter 5, using mathematical modelling, I investigated the effectiveness of pulse
vaccination on susceptible animals, and culling of persistently infected (PI) animals as
interventions to reduce BVDV transmission. The results suggest that the virus reaches
endemic equilibrium at twelve years following introduction into a naïve population, and the
low prevalence of PI animals in the system will continue to maintain transmission in the
absence of intervention. Strategic pulse vaccination, even with limited coverage at 20%, could
limit the spread of BVDV, and coverage approaching 100% significantly reduces BVDV
transmission. Pulse vaccination lowers the prevalence and mitigates disease impact within
smallholder herds. Vaccination combined with culling PI animals would significantly reduce
BVDV prevalence within herds. These provide policymakers with evidence for the best control
methods for the Regions even in the absence of resources to implement systematic control.
The analyses in this thesis highlight the complexity of BVDV circulation and the potential that
(1) there are natural transmissions in Kilimanjaro, and (2) possibly some of the apparent
patterns are due to vaccination in other Regions such as Arusha, Tanga and Iringa. These then
raise critical concerns about vaccine contamination, biosecurity and safety at manufacturing
points. For more successful control, effective vaccination protocols and biosecurity must be
prioritised, and routine vaccine testing for BVDV contamination could be implemented. This
will mitigate transmission and limit the consequences of BVDV such as calf deaths and other
losses due to BVDV in the Tanzanian smallholder dairy system
Essays in health inequality: evidence from 19th century Netherlands
Full text linkPersistent health inequality offers a significant opportunity to improve welfare and
productivity, and understanding its causal pathways is key to adequate policy design. In this
thesis, I first review the literature on health inequality in contemporaneous settings to
establish the contribution of studies using historical data such as the ones provided here.
Studying health shocks in a historical setting can illuminate their potentially different
consequences along the development path. Furthermore, the distribution of improvements to
health conditions within the family may play a key role in the evolution of inequality. Using
death, birth, and marriage certificates of two provinces of the Netherlands, I first analyse the
effect on health inequality of a health shock (an undetermined epidemic) in the early 19th
century. Lower socioeconomic status groups were not only more negatively affected by the
shock but also experienced worse health in the aftermath, worsening overall health inequality
among survivors. I also study the so-called birth-order effect (differences in outcomes
between siblings of different orders of birth) on survivability in the 19th and early 20th
centuries and found a small but significant negative effect of older siblings of the same
gender on survivability and a positive effect of older siblings of different gender. However,
this effect did not increase as life conditions improved, and it was not overly present among
any particular socioeconomic group, suggesting relatively equal access to improvements in
health
Enhancing implicit discourse relation recognition by exploiting label inter-relations
Full text linkImplicit Discourse Relation Recognition (IDRR) is a fundamental yet challenging task
in discourse parsing, as it involves identifying rhetorical, semantic and/or pragmatic
relationships between text spans in the absence of explicit connectives such as “because” or “however”. While recent advances leveraging pre-trained language models
and prompt-based learning have improved performance, most existing approaches treat
discourse sense labels as flat and independent categories. This neglects the rich structural information embedded in annotation frameworks like the Penn Discourse Treebank (PDTB), where discourse relations are organized hierarchically and can co-occur in some ways.
The central claim of this thesis is that structured groupings of discourse senses
— as encoded in the sense hierarchy — can serve as an effective structural prior to
guide model training, particularly by shaping how label distances are represented and
learned. This thesis proposes methods to enhance IDRR by focusing on two kinds
of label inter-relations: the hierarchical relations and co-occurrence-based label interrelations. First, we introduce a contrastive learning framework that utilizes the PDTB
sense hierarchy to guide the selection of semantically meaningful negative examples
during training, thereby encouraging the model to learn finer-grained distinctions between closely related senses. Second, we integrate hierarchical information into a
prompt-based learning paradigm through a prototype-based verbalizer, which aligns
label representations with the sense hierarchy. This approach is further extended to
support zero-shot cross-lingual IDRR, demonstrating effectiveness across both monolingual and cross-lingual scenarios. Third, we explore multi-label classification frameworks to handle cases where multiple discourse relations simultaneously hold between a single pair of text spans — an under-addressed yet prevalent phenomenon in real-world discourse. Incorporating hierarchical sense information also improves the accuracy of multi-label predictions.
Extensive experiments demonstrate the effectiveness of our approaches that consider
the label inter-relations. The results show that explicitly modeling label hierarchies
improves model performance in both single-label classification and multi-label
classification scenarios. This work advances our understanding of how structural relationships between discourse relations can be effectively utilized in computational
models, while also highlighting the importance of handling multi-label cases in discourse
relation recognition.
Finally, this thesis outlines several promising directions for future work. One avenue
is to extend the proposed approaches to broader datasets, including discourse
annotations from alternative frameworks such as RST and eRST, as well as texts from
diverse domains and languages, to better assess the generalizability of the methods.
Another direction involves integrating argument span detection with discourse relation
recognition into a unified framework, thereby advancing toward more realistic and
end-to-end discourse parsing systems. Additionally, future work may explore guiding
Large Language Models (LLMs) to better represent discourse relations and understand
the label relationships between senses by leveraging the hierarchical organization of
discourse senses. Together, these directions point to the broader goal of capturing the
complexity of coherence more effectively in natural language
Structural basis for the inner centromere regulatory hub assembly critical for error-free chromosome segregation
No full textAccurate chromosome segregation during cell division is essential for maintaining genome stability, and errors in this process can result in aneuploidy, polyploidy and disorders such as cancer and infertility. Therefore, understanding the proteins and their interactions at play during this critical event becomes significant. Chromosome segregation is controlled primarily by proteins that are localised at the centromere, the region on the chromosome where kinetochores assemble and attach to microtubules. Within this region, the inner centromere acts as a hub for factors that monitor and control chromosome segregation, including the Chromosomal Passenger Complex (CPC), Heterochromatin Protein 1 (HP1), shugoshins, cohesin, and various kinases and phosphatases that regulate this process. Although these components have been studied extensively, the principles by which they assemble into a functional network remain incompletely understood. This study specifically investigates how the inner centromere interaction network assembles, with the focus on how CPC interacts with H3T3 phosphorylated nucleosome in vitro, and centromeric chromatin in vivo. Using an integrated approach combining biochemical reconstitution, cryo-electron microscopy, cellular localisation and functional assays, and in vitro and in vivo MNase assays, the research characterised the structural and mechanistic basis of CPC association with H3T3ph nucleosome. The cryo-EM structure of CPC-H3T3ph NCP complex revealed how CPC engages with the acidic patch and entry-exit DNA of the NCP via Borealin. EMSA and SPR assays confirmed that the stable and conformationally heterogeneous interactions are important for efficient CPC-NCP binding. Structural findings, supported by biochemical and functional assays, identified a previously unrecognised role of the CPC in protecting nucleosomal DNA against MNase digestion in vitro and in vivo. These results provide mechanistic insight into CPC-mediated chromatin stabilisation and highlight how CPC contributes to maintaining genome integrity. By dissecting the CPC-nucleosome interaction, this study advances the understanding of inner centromere organisation and provides a foundation for further exploration of the assembly and function of the inner centromere interaction network
Human-centred healthcare chatbot design: supporting Chinese overseas students to utilise UK health services
Full text linkLiving in foreign countries can be stressful, particularly in managing health and well-being. Chinese overseas students continuously constitute the largest cohort of international students in the UK; however, they face many challenges in using healthcare services, such as realising the need to access care, communicating with doctors, and understanding treatment processes. This research looked into Chinese overseas students on their behaviours in using UK healthcare services. Based on their pain points, the research project designed and evaluated an educational chatbot prototype that addresses the reluctance issues among the Chinese student population to access healthcare services in the UK. The results showed that the chatbot design boosted Chinese students’ confidence and made them feel more capable of utilising healthcare services
Using the Cerebras CS3 Dataflow Accelerator for HPC
No full textThis project investigates the potential of the Cerebras CS3 wafer-scale accelerator as a platform for computational fluid dynamics (CFD) simulations using the Lattice Boltzmann Method (LBM). The CS3 implementation, written in the Cerebras Software Language (CSL), explores both the opportunities and the practical challenges of deploying a stencil-based, halo-exchange communication pattern on this novel architecture.
The work began with the Taylor–Green Vortex benchmark but pivoted to a fully bounded Poiseuille flow case to reduce boundary-handling complexity. A complete parallel LBM solver was developed, incorporating a fused stream-and-collide kernel and a two-step halo exchange implemented via the low-level color-based communication fabric. Along the way, significant SDK limitations were identified, including restrictions in the high-level message-passing library, stringent local memory constraints, and manual management of low-level hardware resources.
Although a physically correct simulation was not achieved, the project produced a clear characterisation of the failure mode: the breakdown of the no-slip boundary condition, leading to non-physical velocities and chaotic flow patterns.
By scaling the physical parameters appropriately and isolating architectural issues from the numerical model, the instability was traced to a persistent bug within the computational
kernel itself.
The results provide valuable insight into both the current maturity of the Cerebras SDK for non-AI scientific workloads and the development effort required to achieve correctness. This work serves as a detailed case study of the architectural, software, and debugging challenges encountered when porting complex HPC applications to emerging wafer-scale hardware
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!