https://www.um.edu.mt/library/oar/handle/123456789/111128 2026-07-22T20:30:39Z 2026-07-22T20:30:39Z https://www.um.edu.mt/library/oar/handle/123456789/148117 2026-07-20T06:59:05Z 2025-01-01T00:00:00Z

Title: Advancing microRNA target site identification via bias-corrected chimeric datasets for machine learning approaches Abstract: microRNAs (miRNAs) are small, non-coding RNA molecules that regulate gene expression post-transcriptionally. These ∼22-nucleotide long RNA molecules are loaded onto a protein of the Argonaute (AGO) family guiding it to specific RNA target transcripts, which are consequently inhibited from translation to protein. Despite years of research, the precise mechanisms that determine miRNA target recognition remain unclear. Given that a single miRNA can potentially target any RNA transcript, experimental validation of all possible interactions is impractical. To this end, with the recent availability of volumes of data from high-throughput CLASH experiments that capture interacting RNA molecules mediated by a protein, many miRNA target site prediction methods employing data-driven approaches have been developed. However, despite substantial efforts to produce more accurate models, currently, no standardised framework for benchmarking miRNA target site prediction methods exists, and various strategies have been adopted, hindering fair and reliable comparison. Consequently, this undermines the validity of performance claims. Recently, another high-throughput experimental technique, called chimeric eCLIP, was developed, leading to a 70-fold increase in the recovery of miRNA– target site interactions. Here, we identify a unique opportunity to leverage this immense resource of publicly available raw data, to curate benchmark datasets for miRNA target site prediction. Throughout this work, we additionally uncover a miRNA frequency class bias that arises as a result of the method used to generate negative examples in silico. Such methods are required when modeling miRNA target site prediction as a supervised binary classification problem, due to the absence of experimentally confirmed non-binding miRNA–target pairs. To this end, we develop a novel method for generating negatives that mitigates the identified bias. Leveraging data from these high-throughput experiments and applying this new method for generating negatives, we curate three collections of datasets: one novel dataset containing almost three million examples, and bias-corrected versions of two smaller, published datasets. We contribute these datasets to the publicly available and easy-to-use miRBench Python package, providing a framework for benchmarking miRNA target site prediction methods. We benchmark six state-of-the-art deep learning models on these datasets and train simple models to establish a baseline. Retraining a convolutional neural network, that was originally trained on a biased dataset (Average Precision Score (APS): 0.71), on a larger, bias-corrected dataset improves its performance (APS: 0.81), surpassing the previous state of the art (TargetScanCnn, APS: 0.76). This highlights the advantages of well-curated, unbiased benchmarks, facilitating the development of more accurate miRNA target site prediction models thus enabling more reliable downstream analyses. Description: M.Sc.(Melit.)

2025-01-01T00:00:00Z https://www.um.edu.mt/library/oar/handle/123456789/144349 2026-02-27T05:58:41Z 2025-01-01T00:00:00Z

Title: Determining the genotype frequency of the Diego blood group system in Malta Abstract: This study investigated the prevalence of key Diego (Di) blood group antigens (Dia , Dib , Wra , and Wrb ) in Maltese blood donors using a molecular genotyping approach, with the aim of improving rare blood group detection and enhancing transfusion compatibility. A polymerase chain reaction followed by restriction fragment length polymorphism (PCR-RFLP) was developed to genotype four alleles: DI*A (Dia ), DI*B (Dib ), DI02.03* (Wra ), and DI02.04* (Wrb ). High-quality genomic DNA was extracted from donor blood samples, and allele-specific primers were used to amplify target regions of the solute carrier family 4 member 1 SLC4A1 gene. Selected samples were validated through direct DNA sequencing to confirm assay specificity. Additionally, a standardised questionnaire was administered to record donor parental ancestry. The PCR-RFLP method yielded consistent and reproducible results. All individuals tested negative for the rare DI*A and DI02.03* alleles, indicating an absence of the Dia and Wra antigens. The common DI*B and DI02.04* alleles were detected in all samples, suggesting a Di(a-b+) and Wr(a-b+) phenotype across the cohort. Ancestry data showed that the majority of participants reported full Maltese lineage, and allele frequencies were consistent with those expected in European populations. In conclusion, this study provides the first molecular characterisation of the Di Blood Group System in the Maltese population. Although no rare alleles were identified, the developed PCR-RFLP platform proved reliable for Di genotyping and is suitable for integration into routine blood donor screening. These findings establish a reference point for future studies and support the implementation of DNA-based methods to enhance transfusion safety in increasingly diverse populations. Description: M.Sc.(Melit.)

2025-01-01T00:00:00Z https://www.um.edu.mt/library/oar/handle/123456789/143492 2026-02-06T08:14:01Z 2025-01-01T00:00:00Z

Title: Investigating the role of lipopolysaccharide-stimulated monocytes in the development of immune tolerance in community-acquired pneumonia patients Abstract: Introduction: Pneumonia remains a leading cause of morbidity and mortality worldwide. Beyond antibiotic treatment, disease severity hinges on how monocytes balance immune defence and tolerance, a process which may paradoxically worsen outcomes. Background: Emerging evidence links monocyte “tolerance” in CAP to DNA methylation and metabolic shifts, yet why patients differ in cytokine output remains unclear. We hypothesized that transcriptional changes in monocytes underlie this heterogeneity. Methodology: The study included samples from a previously executed prospective observational investigation of CAP patients and control subjects (ELDERBIOME; NCT02928367). Specifically, RNA-seq data of monocytes purified from 75 patients, stimulated with LPS- or not. In addition, TNF-α levels in supernatants were used to stratify samples as LPS responders or non-responders. Read libraries were prepared using the KAPA RNA HyperPrep Kit with RiboErase. Sequencing was performed on the Illumina HiSeq 4000 platform. Bioinformatics included standard quality control metrics, graph-based read alignment, and subsequent DESeq2 modelling, and pathway/network enrichment. Results and Discussion: At padj ≤ 0.01, LPS reshaped expression of 7,033 genes (3,878 upregulated; 3,155 downregulated). Enrichment pinpointed heightened cytokine and interferon pathways, with type I interferons (IFNB1) strongly induced. TNF-α stratification revealed two distinct monocyte states: high responders amplified interferon signalling and HLA class II expression; low responders favoured antioxidant, ECM, and solute transport programmes. Conclusion: These findings uncover transcriptional blueprints explaining patientspecific monocyte behaviour in CAP. Understanding this immune polarity could guide strategies to rebalance hyperinflammation without compromising pathogen clearance. Description: M.Sc.(Melit.)

2025-01-01T00:00:00Z https://www.um.edu.mt/library/oar/handle/123456789/139237 2025-09-25T09:40:47Z 2025-01-01T00:00:00Z

Title: Enhancing the curation of variants of uncertain significance Abstract: The increased demand for genetic sequencing has consequently led to a rise in the number of Variants of Uncertain Significance (VUS), which are classified as such due to insufficient evidence to determine their pathogenicity. These variants lead to incomplete diagnosis for patients, thus potentially causing anxiety and unease due to the uncertainty associated with VUS and may result in unnecessary treatments. The Laboratory of Molecular Pathology and Genetics, Pathology Department, Mater Dei Hospital, situated at the University of Malta, currently utilises software including VarSome and Franklin, to analyse genetic data and provide variant classifications based on the American College of Medical Genetics and Genomics (ACMG) guidelines. Any VUS identified through this analysis are recorded in Excel files which makes management and reclassification challenging. We developed a software, VUSVista, for organised and semi-automated VUS curation by offering ACMG criteria management, referencing of external databases like dbSNP and ClinVar, using standardised Human Phenotype Ontology (HPO) terms to describe sample phenotypes, and linking variants to pseudonymised samples. The absence of personal identifiers in the software safeguards patients’ privacy and makes it GDPR-compliant. VUSVista automatically checks for any updates to ClinVar’s germline classification entries for recorded VUS and searches for newly released publications referencing its VUS using LitVar 2.0. Users receive email notifications whenever a ClinVar entry is updated or a relevant publication is found, which enables them to stay informed about discoveries made by the scientific community. If significant evidence is obtained either from these discoveries or generated internally in the laboratory, users may reclassify the VUS using VUSVista. Our software maintains three audit trails for each of its recorded VUS: one representing any changes applied to the VUS, one for ClinVar update checks and one for publication checks. These audit trails are essential for ensuring traceability and accountability within the diagnostics laboratory’s Quality Management System (QMS). Five individuals working at the aforementioned laboratory participated in one-to-one sessions to evaluate VUSVista and its features. This was followed by a focus group where they discussed the system and shared feedback. During the system’s evaluation, the participants highlighted the automated checks, notification emails, audit trails, and the ability to link variants to samples as significantly useful features that set VUSVista apart from the other systems which they currently use. The participants also suggested improvements to increase the likelihood of adopting the system, such as filtering notifications to only include updates about variants of interest and adding intermediate strengths for ACMG criteria. Overall, VUSVista is a software designed to facilitate VUS curation and reclassification which increases the likelihood of patients receiving a comprehensive diagnosis and in turn, the most accurate management. Description: M.Sc.(Melit.)

2025-01-01T00:00:00Z