1. A generalized empirical interpolation method for direct multi-physics state reconstructionH. Tamim Mahmudul, Francesco A. B. Silva, Rok Krpan, Carlo Fiorina, Jean Ragusa, 2027, izvirni znanstveni članek Povzetek: Reconstructing a coupled multi-physics state from sparse and heterogeneous measurements is central to real-time monitoring and digital twinning, yet it is challenging when only a subset of fields is observable and sensors operate over field-dependent regions. This work introduces the Multi-field Generalized Empirical Interpolation Method, which extends the Generalized Empirical Interpolation Method to product spaces by treating the full coupled state as a single element of a multi-field Hilbert space while allowing measurements to be selected across multiple fields and sensing modalities. In the offline phase, the Multi-field Generalized Empirical Interpolation Method constructs a reduced basis and a corresponding set of measurement functionals through a greedy procedure that (i) simultaneously identifies the global basis function and the field to be sensed and (ii) improves numerical robustness by applying an explicit scaling factor to each selected measurement functional. In the online phase, the method reconstructs all fields, including unmeasured ones, by solving a small interpolation system from noisy measurements. Global and field-wise stability measures (Lebesgue constants) and trace-based noise-amplification indicators are also introduced to provide an exact characterization of the expected mean-square contribution of Gaussian measurement perturbations. Numerical experiments on a two-dimensional molten salt reactor benchmark demonstrate accurate reconstruction under realistic observability constraints and quantify the trade-off between reduced-space approximation and noise sensitivity.
Ključne besede: multi-physics state reconstruction, generalized empirical interpolation method, reduced-order modeling, sensor placement, data assimilation, digital twinning, nuclear reactor monitoring
Objavljeno v DiRROS: 23.06.2026; Ogledov: 10; Prenosov: 8
 Povezava na datoteko |
2. Early warning systems for chemical risks in Europe : methodological components for signal detection, prioritization, uncertainty management and follow-upAchilleas Karakoltzidis, Nikiforos Alygizakis, Patrik L. Andersson, Tina Kosjek, Lutz Ahrens, 2026, pregledni znanstveni članek Povzetek: Chemical pollution can affect ecosystems and human health, highlighting the need for approaches that identify, evaluate, and prioritize emerging chemical risks before they become established public-health issues requiring regulatory actions. This review and perspective article examines methodological components required for Early Warning Systems (EWSs) for chemical risks, with particular attention to the European policy context. It examines methodological components for chemical EWSs rather than proposing a fully implemented system. It considers how state-of-the-art and emerging methods can support signal generation, signal strengthening, prioritization, uncertainty assessment, communication, and follow-up. The reviewed components include matrices and sampling strategies; chemical monitoring, suspect screening, and non-target screening; effect-based methods, New Approach Methodologies, and effect-directed analysis; exposure and hazard modelling; QSAR, read-across, AI-supported and data-mining tools; expert evaluation; and governance processes that link scientific signals to proportionate follow-up. The conceptual workflow is used as an example of how these components may be organized into a signal-handling process, while EU-level developments provide the broader policy and governance context. The actionable value of a chemical EWS does not depend on any single method, but on structured integration of complementary evidence streams. An effective EWS should combine sensitive weak-signal detection with transparent prioritization, explicit uncertainty assessment, FAIR and interoperable data infrastructures, and clearly assigned responsibilities for communication and follow-up.
Ključne besede: real-time data, early warning system, hazard prioritization, chemical risks
Objavljeno v DiRROS: 15.06.2026; Ogledov: 122; Prenosov: 74
 Celotno besedilo (1,25 MB)
Gradivo ima več datotek! Več... |
3. Control compounds for preclinical drug-induced liver injury assessment : consensus-driven systematic review by the ProEuroDILI networkAntonio Segovia-Zafra, Marina Villanueva-Paz, Ana Sofia Serras, Gonzalo Matilla-Cabello, Ana Bodoque-Garcia, Daniel E. Di Zeo-Sánchez, Hao Niu, Ismael Alvarez-Alvarez, Laura Sanz-Villanueva, Sergej Godec, Irina Milisav, 2024, izvirni znanstveni članek Povzetek: Background & aims: Idiosyncratic drug-induced liver injury (DILI) is a complex and unpredictable event caused by drugs, and herbal or dietary supplements. Early identification of human hepatotoxicity at preclinical stages remains a major challenge, in which the selection of validated in vitro systems and test drugs has a significant impact. In this systematic review, we analyzed the compounds used in hepatotoxicity assays and established a list of DILI-positive and -negative control drugs for validation of in vitro models of DILI, supported by literature and clinical evidence and endorsed by an expert committee from the COST Action ProEuroDILI Network (CA17112). Methods: Following 2020 PRISMA guidelines, original research articles focusing on DILI which used in vitro human models and performed at least one hepatotoxicity assay with positive and negative control compounds, were included. Bias of the studies was assessed by a modified 'Toxicological Data Reliability Assessment Tool'. Results: A total of 51 studies (out of 2,936) met the inclusion criteria, with 30 categorized as reliable without restrictions. Although there was a broad consensus on positive compounds, the selection of negative compounds lacked clarity. 2D monoculture, short exposure times and cytotoxicity endpoints were the most tested, although there was no consensus on drug concentrations. Conclusions: Extensive analysis highlighted the lack of agreement on control compounds for in vitro DILI assessment. Following comprehensive in vitro and clinical data analysis together with input from the expert committee, an evidence-based consensus-driven list of 10 positive and negative control drugs for validation of in vitro models of DILI is proposed. Impact and implications: Prediction of human toxicity early in the drug development process remains a major challenge, necessitating the development of more physiologically relevant liver models and careful selection of drug-induced liver injury (DILI)-positive and -negative control drugs to better predict the risk of DILI associated with new drug candidates. Thus, this systematic study has crucial implications for standardizing the validation of new in vitro models of DILI. By establishing a consensus-driven list of positive and negative control drugs, the study provides a scientifically justified framework for enhancing the consistency of preclinical testing, thereby addressing a significant challenge in early hepatotoxicity identification. Practically, these findings can guide researchers in evaluating safety profiles of new drugs, refining in vitro models, and informing regulatory agencies on potential improvements to regulatory guidelines, ensuring a more systematic and efficient approach to drug safety assessment.
Ključne besede: clinical data, control compounds, drug-induced liver injury, expert committee, panel of control drugs, preclinical drug safety testing, validation of in vitro DILI models
Objavljeno v DiRROS: 11.06.2026; Ogledov: 112; Prenosov: 96
Celotno besedilo (1,10 MB)
Gradivo ima več datotek! Več... |
4. Borrelia PeptideAtlas : a proteome resource of common Borrelia burgdorferi isolates for Lyme researchPanga J. Reddy, Zhi Sun, Helisa H. Wippel, David H. Baxter, Kristian Swearingen, David D. Shteynberg, Mukul K. Midha, Melissa J. Caimano, Klemen Strle, Yongwook Choi, 2024, izvirni znanstveni članek Povzetek: Lyme disease is caused by an infection with the spirochete Borrelia burgdorferi, and is the most common vector-borne disease in North America. B. burgdorferi isolates harbor extensive genomic and proteomic variability and further comparison of isolates is key to understanding the infectivity of the spirochetes and biological impacts of identified sequence variants. Here, we applied both transcriptome analysis and mass spectrometry-based proteomics to assemble peptide datasets of B. burgdorferi laboratory isolates B31, MM1, and the infective isolate B31-5A4, to provide a publicly available Borrelia PeptideAtlas. Included are total proteome, secretome, and membrane proteome identifications of the individual isolates. Proteomic data collected from 35 different experiment datasets, totaling 386 mass spectrometry runs, have identified 81,967 distinct peptides, which map to 1,113 proteins. The Borrelia PeptideAtlas covers 86% of the total B31 proteome of 1,291 protein sequences. The Borrelia PeptideAtlas is an extensible comprehensive peptide repository with proteomic information from B. burgdorferi isolates useful for Lyme disease research.
Ključne besede: Borrelia burgdorferi, Lyme disease, peptide repository, proteomic data
Objavljeno v DiRROS: 02.06.2026; Ogledov: 119; Prenosov: 83
Celotno besedilo (4,63 MB)
Gradivo ima več datotek! Več... |
5. Arabidopsis tissue- and perturbation-specific gene expression resource : version v2Vid Modic, Jan Zrimec, 2026, zaključena znanstvena zbirka raziskovalnih podatkov Povzetek: A prerequisite to understanding how an organism functions and responds to its environment is to determine which gene expression patterns are associated with a specific tissue type or perturbation response. Neural network-based methods can provide subsets of highly informative genes for such associations. Here, we propose that integrating prior molecular knowledge related to gene expression within deep neural networks can lead to improved identification of tissue and perturbation-related gene sets. We first construct an Arabidopsis tissue- and perturbation-specific gene expression resource from published data, and address batch effects by implementing and evaluating several approaches, including Conditional Variational Autoencoders. We then incorporate prior published molecular knowledge on protein-DNA and protein-protein interactions as additional model layers, training the models to classify tissue types and perturbation groups according to input gene expression patterns. Although knowledge graph-based models achieve similar classification performance as baseline models, the analysis of model explainability demonstrates that they outperform the baseline models by prioritising biologically relevant genes. The identified genes are shown to be related to the specific tissue types and molecular processes following the particular perturbations. Our results demonstrate the applicability and reliability improvements of knowledge graph-primed deep learning for identifying condition-specific genes and gene sets.
Ključne besede: transcriptomic data, bioinformatics, Arabidopsis
Objavljeno v DiRROS: 22.05.2026; Ogledov: 104; Prenosov: 144
 Raziskovalni podatki (1,24 GB)
Gradivo ima več datotek! Več... |
6. |
7. |
8. Data and data quality in mathematicsKatja Berčič, 2026, samostojni znanstveni sestavek ali poglavje v monografski publikaciji Povzetek: Pure mathematics is often viewed, even by its practitioners, as a discipline in which data play little or no role. Data, when acknowledged at all, are often seen as a byproduct of research rather than a research product in their own right. Yet databases and datasets are increasingly central to the way mathematicians formulate conjectures, test hypotheses, and explore complex structures. Unlike empirical data, data in mathematics often consist of exact values derived from symbolic definitions or computations and commonly describe highly structured objects such as graphs, elliptic curves, or manifolds. This combination of abstraction, precision, and low redundancy poses distinctive challenges for data quality, shifting the focus away from concerns like noise and bias toward correctness, completeness, consistency, and accessibility.
Ključne besede: mathematical knowledge management, digital mathematics libraries and repositories, computer-assisted mathematics, implementation challenges, data quality dimensions, mathematical data
Objavljeno v DiRROS: 06.05.2026; Ogledov: 162; Prenosov: 153
Celotno besedilo (831,18 KB)
Gradivo ima več datotek! Več... |
9. |
10. Fake news detection through LLM-driven text augmentation across media and languagesAbdul Sittar, Mateja Smiljanić, Alenka Guček, Marko Grobelnik, 2026, izvirni znanstveni članek Povzetek: The proliferation of fake news across social media, headlines, and news articles poses major challenges for automated detection, particularly in multilingual and cross-media settings affected by data imbalance. We propose a fake news detection framework based on LLM-driven, feature-guided text augmentation. The method generates realistic synthetic samples across languages, media types, and text granularities while preserving mean ing and stylistic coherence. Experiments with classical and transformer-based models (Random Forest, Logistic Regression, BERT, XLM-R) across social media, headlines, and multilingual news datasets show consistent improvements in performance. For inherently balanced datasets (e.g., social media), synthetic augmentation yields negligible but stable performance changes. Across imbalanced scenarios, synthetic augmentation substantially improves minority-class recall and F1-score (e.g., fake news recall from 0.57 to 0.86), while preserving majority-class performance, leading to more balanced and reliable classifiers, whereas oversampling significantly degrades results due to overfitting on duplicated language patterns. Overall, a hybrid semantic- and style-based model proves to be the most robust strategy, outperforming oversampling and matching or exceeding baseline performance across datasets
Ključne besede: fake news detection, low-resource languages, data imbalance, synthetic data generation, prompt engineering, style-based features, semantic features
Objavljeno v DiRROS: 28.04.2026; Ogledov: 247; Prenosov: 165
Celotno besedilo (1,16 MB)
Gradivo ima več datotek! Več... |
