179566 research outputs found
Sort by
Valid Conformal Prediction for Dynamic GNNs
Get PDFDynamic graphs provide a flexible data abstraction for modelling many sorts of real-world systems, such as transport, trade, and social networks. Graph neural networks (GNNs) are powerful tools allowing for different kinds of prediction and inference on these systems, but getting a handle on uncertainty, especially in dynamic settings, is a challenging problem. In this work we propose to use a dynamic graph representation known in the tensor literature as the unfolding, to achieve valid prediction sets via conformal prediction. This representation, a simple graph, can be input to any standard GNN and does not require any modification to existing GNN architectures or conformal prediction routines. One of our key contributions is a careful mathematical consideration of the different inference scenarios which can arise in a dynamic graph modelling context. For a range of practically relevant cases, we obtain valid prediction sets with almost no assumptions, even dispensing with exchangeability. In a more challenging scenario, which we call the semi-inductive regime, we achieve valid prediction under stronger assumptions, akin to stationarity. We provide real data examples demonstrating validity, showing improved accuracy over baselines, and sign-posting different failure modes which can occur when those assumptions are violated
Comprehensive mapping of the 5′ and 3′ untranslated regions of Aspergillus fumigatus reveals diverse mechanisms of mRNA processing including premature transcription termination.
Get PDFIn the twenty years since the first genome sequencing of Aspergillus fumigatus, the field has seen an explosion in both the number of sequenced genomes and our molecular understanding of this ubiquitous human fungal pathogen. Despite an improved knowledge of the A. fumigatus genome, we still know little about the transcriptome, with key regulatory sequences like the untranslated regions of mRNA based only on in silico predictions and bulk-RNA-seq. Here, we provide an improved description of 5′ and 3′ untranslated regions of A. fumigatus poly(A)-enriched RNA through experimental mapping of transcription start sites and polyadenylation sites using 5′ and 3′ End-Seq. We assigned high-quality 5′ ends to 2,747 genes (average length 126 nt), 3′ ends to 7,079 genes (average length 268 nt), and improved our understanding of the regulatory landscape of A. fumigatus gene expression. We leveraged the refined 5′ UTRs to identify upstream open reading frames and binding sites for important RNA binding proteins like the translational regulator Ssd1 and the 3′ UTRs to define binding sites for PUF proteins known to contribute to mRNA localization and regulation. Although a single isoform typically dominated expression, we observed 148 instances of alternative start sites and 1,675 alternative stop sites. Interestingly, we detected multiple examples of premature transcriptional termination, including the first evidence for promoter-proximal premature transcriptional termination in a member of the Eurotiomycetes. Ultimately, we provide a resource to the Aspergillus community and an accurate starting point for unravelling the complexities of gene regulation in an important human pathogen
Evidence for causal links between known modifiable risk factors and dementia: A systematic review of Mendelian randomisation studies
Get PDFBackgroundWe aimed to systematically review the evidence for associations between the known modifiable risk factors and dementia based on Mendelian randomisation (MR) studies.MethodFive databases were searched from inception to April 2024 investigating the association between the 12 risk factors identified in the Lancet Commission and dementia. Evaluable analyses were categorized into one of four levels (robust, probable, suggestive, insufficient) based on estimate significance level and concordance of direction of effect between main and sensitivity analyses. Evidence from clinically diagnosed dementia outcomes was synthesized separately from proxy outcomes. A post hoc sensitivity analysis excluded estimates with concerns over construct validity.ResultsA total of 47 studies were included, representing 240 MR associations (185 unique and evaluable). Over half (73.5%) of evaluable analyses were graded as providing insufficient evidence for a causal association. Among clinically diagnosed outcomes, the strongest evidence was for educational attainment (mainly probable evidence in a protective direction) and type 2 diabetes-related dysfunction (probable evidence in the risk direction). Smoking showed probable evidence of a protective association. Other risk factors, produced inconclusive or insufficient evidence. Proxy outcome analyses yielded weaker findings; in particular, the association between education and Alzheimer's disease reversed direction.ConclusionMR evidence for most Lancet Commission risk factors remains insufficient or inconclusive. The most consistent support for causal associations was observed for lower educational attainment and type 2 diabetes. Null findings should be interpreted cautiously given limitations in GWAS phenotyping, sample composition, and MR methodology.<br/
“It’s Their Little Red Bible”:Exploring immigration and immunization journeys through Nigerian mother relationships with child health records in London
Get PDFDrawing on an ethnography of Nigerian migrant motherhood in London, we explore how the original design of children’s health records, known as the Red Book, to log routine vaccinations has been transformed to document family proof of residency amidst Britain’s “hostile environment.” Nigerian migrant mothers use the Red Book as evidence of raising British children and to overcome their precarity. By resolutely documenting their children’s immunizations, Nigerian mothers pursue both biological and political immunity while on a simultaneous immunization and citizenship journey
Invoking security to bypass procedure:The European Union's Critical Medicines Act
Get PDFThe European Union's recently proposed Critical Medicines Act (CMA) was published without recourse to standard democratic policy-making procedures. Framed by the European Commission as an urgent response to a pressing security threat, the CMA was not subject to an impact assessment, and the stakeholder consultation designed to inform its development was both short and last-minute. In this commentary, we examine the implications of the CMA case for the democratic legitimacy of EU policy-making. We argue that, since medicine shortages have been on the EU agenda for nearly a decade and the CMA extends beyond emergency provisions to address long-term industrial policy, it does not constitute an urgent or exceptional policy issue in the usual sense. Situating the CMA within wider patterns of EU health securitisation, we highlight the risks posed by circumventing procedural safeguards, concluding that this shift may exacerbate perceptions of a democratic deficit within EU governance.</p
Mapping of Nab3 RNA-binding sites in Saccharomyces cerevisiae with abrogated binding of Nab3 to PIC2
No full textUsing CRAC, we compared the transcriptomic occupancy of Nab3 in a Saccharomyces cerevisiae BY4741 parental strain (PIC2-GFP) and two derived mutants lacking Nab3 RNA-binding sites in PIC2. The purpose of the experiment was two-fold: on the one hand, we aimed to verify that the mutations inserted in the Nab3 binding sequences of PIC2 had indeed abrogated binding of Nab3 to the PIC2 transcript; on the other hand, we wanted to check whether differential binding of Nab3 to PIC2 would affect how the protein bound other targets in the genome. Sequencing outputs were processed using the pyCRAC pipeline and peak calling was performed with DBPeaks, our newly developed package for identification and comparison of binding sites defined by RNA-binding footprinting techniques (e.g., CRAC, iCLIP, PAR-CLIP, etc.)
Mapping of Nab3 RNA-binding sites in Saccharomyces cerevisiae with suboptimal expression of PIC2.
No full textUsing CRAC, we compared the transcriptomic occupancy of Nab3 in a Saccharomyces cerevisiae BY4741 parental strain (PIC2-GFP, i.e., WT) and two derived mutants overexpressing (pTEF1-PIC2) or lacking (KO) PIC2. The purpose of the experiment was to check whether overexpressing or preventing the expression of PIC2 (an established Nab3 mRNA target which, when overexpressed or deleted, causes severe cellular defects) would cause a re-distribution of Nab3 binding among its other target transcripts. Sequencing outputs were processed using the pyCRAC pipeline and peak calling was performed with DBPeaks, our newly developed package for identification and comparison of binding sites defined by RNA-binding footprinting techniques (e.g., CRAC, iCLIP, PAR-CLIP, etc.)
A qualitative study of the experiences of victim-survivors of TA-CSA around accessing and receiving support from professionals and relevant services
No full text## Access ## This dataset is held in the Edinburgh DataVault, directly accessible only to authorised University of Edinburgh staff. External users may request access to a copy of the data by contacting the Principal Investigator, Contact Person or Data Manager named on this page. Requests for access will not necessarily be granted. University of Edinburgh users who wish to have direct access should consult the information about retrieving data from the DataVault at: https://www.ed.ac.uk/is/research-support/datavault.This is the archived dataset from Anna Balmer's DClinPsychol research project (completed in September 2025), supervised by Dr Juliane Kloess. The dataset is comprised of ten interview transcripts and MS Word files recording the different stages of analysis. The data files are being archived for research integrity purposes, and are not available or accessible to anyone outside of the research team
Access Repository Dashboard
Do you manage Edinburgh Research Explorer? Access insider analytics, issue reports and manage access to outputs from your repository in the CORE Repository Dashboard!