Abstract: Susceptibility to misinformation threatens democratic processes, yet the differential predictive role of authoritarianism subdimensions remains understudied. This study integrates data from two independent Romanian samples collected during the 2024–2025 electoral cycle (N = 597), examining how aggression, conventionalism/traditionalism, and submission/conservatism predict two distinct outcomes: accuracy in identifying false news (Study 1, N = 427, RWA-15) and belief in conspiracy statements drawn from the annulled Romanian presidential election (Study 2, N = 170, VSA-6). Hierarchical regressions revealed a striking dissociation: authoritarian submission was the dominant predictor of fake news detection deficits (β = −0.41, p < .001), while traditionalism was the dominant predictor of conspiracy belief (β = +0.39, p < .001). Aggression failed to predict either outcome after controls. Fisher's z tests confirmed systematic cross-study differences (both p < .001), consistent with distinct cognitive and ideological routes to misinformation susceptibility. Person-centered analyses replicated the variable-centered findings. Cross-instrumental convergence was partial: the traditionalism/conventionalism equivalence replicated consistently, while submission/conservatism showed less stability. These findings suggest that prebunking interventions should be tailored to authoritarian profiles: cognitively oriented training for submission-dominant individuals, and value-based argumentation for traditionalism-dominant ones.

Abstract: Importance The increase in prevalence of autistic spectrum disorder (ASD) in children in most countries of the world has been attributed to assortative mating meaning that spouses have found each other guided by features of ASD leading to increased genetic predisposition in the offspring. This does not explain the differences in prevalence of ASD in subgroups of the population with similar mating patterns. Objective: To test the hypothesis that parental genetic heterogeneity is associated with ASD in the offspring.Design: In this comparative observational study data were included from population based studies reporting number of patients with ASD in populations with known high, moderate or low genetic heterogeneity. The prevalence (per 10000 of population) was compared between groups. Compared were the percentages of children with learning difficulties in the three groups.Setting: All countries with data from investigations reporting prevalence of ASD in ethnic subgroups of the population.Participants: Children with and without ASD of parents with a known degree of genetic heterogeneity from population based studies allowing analysis of individual participant data.Exposures: Based on pre-existing data of genetic and anthropological studies subgroups of childhood populations with low, moderate and high parental genetic heterogeneity were investigated.Main outcome and measures: Prevalence of ASD and percentage of children with ASD and learning difficulties in offspring of parents with different degrees of genetic heterogeneity were compared. The hypothesis that parental genetic heterogeneity is associated with ASD in offspring was formulated prior to the study as a result of data of a previous study and stated there.Results: The prevalence of ASD across spouse pairs of high, moderate or low genetic heterogeneity for 182016 children with ASD in childhood populations totalling 62290632 children in the USA, the United Kingdom, Sweden, Israel and Australia were compared. In all studies reporting data prevalence of ASD was significantly higher in offspring of parents with high genetic heterogeneity ranging from 12.9 to 280.5 compared to offspring of parents with moderate genetic heterogeneity with a range of 9.0 to 243.1 and low genetic heterogeneity with a range of 4.6 to 71.7. The percentage of children with learning difficulties showed no consistent difference between groups.Conclusion and relevance: Autism spectrum disorder in offspring is associated with parental genetic heterogeneity. This will direct future research towards gene constellations protective against expression of ASD predisposing genes.

Abstract: Veterinarians are occupationally exposed to zoonotic pathogens and may therefore be at increased risk of vaccine-preventable diseases. However, evidence on vaccination coverage and immunization practices in this professional group remains limited. This study assessed vaccination coverage, perceptions, and barriers to immunization among registered veterinarians in Málaga, Spain, within a One Health framework. A cross-sectional anonymous survey was distributed to all registered veterinarians in the province. The questionnaire collected sociodemographic and professional information, self-reported vaccination status, attitudes toward vaccination, and training needs. A total of 164 veterinarians participated. Vaccination coverage was low for several occupationally relevant vaccines, including seasonal influenza (21.3%) and rabies (23.2%). Nearly half of respondents were unaware of their hepatitis B vaccination status. In contrast, COVID-19 vaccine uptake was high (96.9%). Although most participants considered vaccination highly important, a marked gap between positive attitudes and actual vaccination practices was identified. These findings highlight important gaps in occupational immunization among veterinarians and support the need for improved vaccination strategies, easier access to immunization programs, and enhanced vaccinology training. Strengthening vaccination coverage among veterinary professionals may also contribute to zoonotic disease prevention and preparedness at the human–animal interface within a One Health approach.

Abstract: This study presents a field-based, empirically grounded investigation into the spatiotemporal dynamics and indoor–outdoor (I/O) coupling mechanisms of PM₂.₅ in residential buildings across North China. Concurrent high-resolution (10-min interval) measurements of indoor and outdoor PM₂.₅ mass concentrations were conducted from April to December 2025 across six instrumented residential units—stratified by urban/rural setting, building age, heating infrastructure, and envelope integrity—to capture representative heterogeneity in exposure contexts. Data were acquired using calibrated β-attenuation monitors (BAM-1020, ±2.5 µg/m³ accuracy) with integrated temperature/humidity compensation, and synchronized via GPS time-stamping to ensure temporal alignment.Statistical analysis employed rigorous inferential methods: paired t-tests (α = 0.001, two-tailed) confirmed statistically significant concentration disparities between indoor and outdoor environments (p < 0.001 for all sites), rejecting the null hypothesis of I/O equilibrium. Linear regression modeling (R², slope, intercept, residual diagnostics) quantified infiltration-driven coupling strength, while Pearson correlation coefficients (r) and associated p-values assessed monotonic dependence under varying operational conditions. The observed I/O ratio spanned 0.674–2.673, reflecting pronounced building-specific modulation of infiltration and source dominance. Critically, under window-open conditions, residences with active indoor sources (e.g., cooking, incense burning, biomass space heating) exhibited mean I/O ratios >1.0 (1.32 ± 0.18), whereas under window-closed conditions—where infiltration is suppressed—the same units registered I/O <1.0 (0.87 ± 0.11), indicating net indoor generation outweighing penetration loss. In contrast, source-free dwellings maintained strong linear I/O correlation (r = 0.89–0.95, p < 0.001) across both ambient and haze episodes (PM₂.₅ > 150 µg/m³), with regression slopes consistent with empirically derived infiltration factors (0.62–0.78). Conversely, source-active units displayed statistically significant negative Pearson correlations (r = −0.41 to −0.63, p < 0.001) during source events—demonstrating dynamic decoupling wherein indoor concentrations diverge inversely from outdoor trends due to dominant internal emission fluxes.

Abstract: Event logs record enterprise service processes as sequences of timestamped activities. Predictive process monitoring typically trains a separate model per event log, requiring retraining when processes, activity vocabularies, or time scales change.We study an event-log-native foundation model pretrained once on heterogeneous logs and adapted to an unseen log with a small labeled support set, without fitting a separate model for that log. The model represents prefixes with a mixture-of-experts transformer and predicts next activity and remaining time with a support-conditioned prototypical head whose label space is defined by the support context.To align training with deployment, we incorporate retrieval-centric objectives that shape the representation for nearest-neighbor support selection and we provide confidence estimates for both classification and regression.We benchmark this no-fine-tuning setting on five public logs against classical CPU-friendly baselines and a general-purpose in-context tabular predictor, down to 0.5% of training cases. Results show that one pretrained predictor can be competitive, but performance depends on retrieving suitable supports; in several settings, simple kNN on learned embeddings matches the full head.We also find that activity-time dependence is informative in this benchmark, and that confidence scores support useful performance stratification and expert-level representation analysis.

Abstract: Background/Objectives: Participation in sport and daily activities among individuals with spinal cord injury (SCI) is influenced not only by physical capacity but also by psychological factors such as balance confidence. Existing measures primarily assess performance or fear of falling, leaving a gap in evaluating perceived sitting balance ability. This study aimed to (1) develop a Patient-Reported Sitting Balance Confidence (PR-SBC) scale for non-ambulatory individuals with SCI, (2) incorporate participation-based activities relevant to community and sport, and (3) evaluate its psychometric properties. Methods: Scale development used a Delphi process with clinicians, researchers, individuals with SCI, and care partners to refine items. The final 19-item PR-SBC was administered to 31 individuals with SCI (inpatient and outpatient). Inter-rater reliability was assessed using intraclass correlation coefficients (ICC), internal consistency via Cronbach’s alpha, and criterion validity through correlation with the SCI Falls Concern Scale (SCI-FCS). Exploratory regression analyses examined demographic and clinical predictors of PR-SBC scores. Results: The PR-SBC demonstrated excellent inter-rater reliability (ICC = 0.953) and strong internal consistency (α = 0.931). Criterion-related validity showed a moderate inverse correlation with SCI-FCS scores (r = −0.512). Outpatient setting and manual wheelchair use were associated with higher balance confidence, while injury-related variables were not significant predictors. Conclusions: The PR-SBC is a reliable and valid instrument for assessing sitting balance confidence in individuals with SCI. By capturing perceived balance across functional and participation contexts, it complements existing measures and may support improved clinical decision-making, targeted interventions, and enhanced participation outcomes.

Abstract:

Long-context language model inference is increasingly limited by the memory footprint and bandwidth cost of KV caches, especially when retrieval-heavy prompts require preserving information across many layers. We introduce DepthSketch-KV, a cache compression method that exploits cross-layer redundancy by storing a shared low-rank residual sketch for adjacent transformer layers together with small per-layer key and value corrections. During decoding, query-conditioned depth gates estimate attention drift and choose whether each layer should access its local full-rank cache, the shared compressed cache, or a mixed reconstruction. A calibration-free error controller converts logit drift into per-sequence cache budgets, avoiding task-specific tuning while adapting to heterogeneous prompts. Across LongBench, Needle-in-a-Haystack, NarrativeQA, Qasper, HotpotQA, GovReport, MultiNews, HumanEval, and MBPP, DepthSketch-KV consistently reduces KV cache memory at 32k and 64k contexts while preserving retrieval, QA, summarization, and code completion accuracy. Compared with MiniCache, H2O, StreamingLLM, SnapKV, PyramidKV, KIVI, and FlexGen, it improves the accuracy-memory tradeoff, increases decoding throughput, and reduces time to first token with only small changes in perplexity.

Abstract: Background: Self-care and healthy dietary habits are fundamental components of heart failure (HF) management, yet the relationship between dietary quality and self-care remains insufficiently explored. This study investigated the association between dietary habits and self-care behaviors among patients with HF. Methods: A cross-sectional study was conducted among 137 adults with HF attending outpatient clinics in Greece. Self-care behaviors were assessed using the Hippocratic Heart Failure Self-Care Scale (HHFSCS), whereas dietary quality was evaluated using the validated Greek version of the Cardiovascular Diet Questionnaire-2 (CDQ-2). Associations between demographic characteristics, dietary quality, and self-care were initially examined using non-parametric tests. Two multivariable General Linear Models (GLMs) were subsequently performed to identify independent predictors of self-care. Results: The mean HHFSCS score was 41.0 ± 11.3, indicating a moderate level of self-care. Medication adherence and attendance at scheduled appointments showed the highest scores, whereas exercise, smoking-related behaviors, alcohol-related behaviors, and symptom monitoring demonstrated lower adherence. The mean total CDQ-2 score was −5.9 ± 3.9. In the multivariable analysis, overall dietary quality remained independently associated with self-care after adjustment for age, sex, marital status, educational level, and employment status (β = −0.511, p = 0.020). Among the individual dietary domains, only saturated fat intake remained independently associated with self-care (p = 0.015), whereas fruit consumption and unsaturated fatty acid intake were not significant predictors. Employment status also remained independently associated with self-care (p < 0.001). Conclusions: Better dietary quality, particularly lower saturated fat intake, was independently associated with improved self-care among patients with HF. These findings support integrating comprehensive nutritional assessment and individualized dietary counseling into routine HF management. Dietary behavior may serve as a practical marker of patients' overall self-management capacity and could assist healthcare professionals in identifying individuals who require additional educational and behavioral support.

Abstract: The rapid proliferation of digital pathology has created an urgent need for integrated, scalable, and secure platforms capable of supporting the full lifecycle of whole-slide image (WSI) analysis — from quality assessment through collaborative annotation to AI model deployment. Existing tools address these requirements in isolation, creating fragmented workflows that impede clinical adoption and AI development. Here we present A!Path, a modular ecosystem comprising three synergistic components: (1) A!magQC, a fully automated AI-assisted quality control pipeline assessing five image quality metrics across H&E and multiplex fluorescence modalities; (2) A!HistoClouds, a unified annotation and inference platform that combines cloud-based scalability with local server flexibility, implementing a three-phase closed-loop pathologist-AI interaction workflows, Segment Anything Model (SAM)-based annotation, multi-user collaborative workflows, integrated AI inference (A!Prostate), and project analytics on a deployable backend (A!Server); and (3) A!Secure, a cryptographic security layer for WSI protection, providing application-layer encryption, policy-based access control, streaming tile decryption, and format-aware protection of compressed, pyramid-structured pathology images. A!HistoClouds is deployable on both cloud infrastructure and institutional local servers, enabling organizations to select the deployment model best suited to their data governance requirements without sacrificing platform capability. Validation across a cohort of 302 prostate tissue specimens demonstrated that A!magQC achieved greater than 95% agreement with expert visual quality assessment, while A!Secure-protected WSI regions yielded a low PSNR of 7.42 dB and SSIM of 0.0289 relative to the original images, indicating strong visual obfuscation of diagnostically relevant content. A!Path provides a coherent, deployment-flexible foundation for clinical diagnostic AI development and multi-centre digital pathology collaboration.

Abstract: Smart City research has advanced significantly during the last two decades; however, many existing frameworks remain fragmented in their treatment of technology, governance, sustainability, regional development and quality of life. This study develops the N-SCIM Framework (Nikolov Smart City Integrated Framework) as an integrated conceptual architecture for smart city and regional ecosystem governance. The paper applies a qualitative theoretical-applied methodology based on systematic literature review, comparative conceptual analysis and conceptual synthesis. The results establish the theoretical foundations, definition, core principles, dimensions and comparative positioning of the N-SCIM Framework. The framework integrates smart governance, smart economy, smart people, smart environment, smart mobility and infrastructure, and smart technologies and data within a unified smart regional ecosystem. Unlike traditional models, N-SCIM positions quality of life, sustainability and adaptability as the central outcomes of territorial intelligence. The study contributes to smart city theory by extending the analytical perspective from isolated urban systems toward interconnected regional ecosystems and by incorporating AI-enabled and data-driven governance as structural components of contemporary territorial management. Future research should operationalize the framework through indicators, the N-Score composite index and empirical validation across different urban and regional contexts.

Abstract: This article investigates generalized trans-Sasakian manifolds and their rich geometric structure. We present a method for constructing two additional almost contact metric structures from a given generalized trans-Sasakian manifold. A complete classification of these structures is obtained through the study of their integrability and normality properties. We further examine these structures when they are realized as real hypersurfaces of complex space forms. Several existence and non-existence results are established.

Abstract: Distributed generation offers a critical pathway to sustainable urban energy; however, in-complete regulatory frameworks in emerging economies frequently force residential pho-tovoltaic (PV) systems into export-restricted operation. This study quantifies the empirical performance and opportunity costs of a demand-matched, zero-export PV system paired with an Internet of Things (IoT)-managed backup tier. A 54-month longitudinal sin-gle-case evaluation was conducted in Arequipa, Peru, utilizing a 1.36 kWp grid-tied array and a 0.6 kWh/day decoupled backup subsystem. Performance was rigorously bench-marked against a fully metered pre-PV counterfactual. The zero-export constraint man-dated a 96.3% self-consumption index (SCI), capping the self-sufficiency index (SSI) at 38.7% and yielding a utilization ratio (U) of only 24.1% against the site's technical poten-tial. Economically, the status quo produced a negative net present value (NPV) of −USD 885. However, scenario modeling demonstrated that activating a net-billing or net-metering framework robustly reversed the NPV to +USD 792 and +USD 4,587, respec-tively. Furthermore, the decoupled backup architecture successfully mitigated 97.7% of 43 recorded grid outages at a resilience cost of USD 3.2 per protected hour. The headline finding is therefore institutional, not technological: identical hardware shifts from val-ue-destroying to highly profitable purely through the regulatory mechanism governing surplus energy. We present two replicable sizing rules and propose the U metric to trans-late regulatory impasse into an internationally standardized reporting parameter, provid-ing an evidence-based roadmap for accelerating decentralized renewable energy in re-source-constrained grids (SDG 7, SDG 11).

Abstract: Diabetic foot ulcers (DFUs) are a leading cause of lower-extremity amputation world wide, yet routine assessment still relies on manual visual inspection by trained clinicians, a resource that is subjective and unevenly available in low-resource settings. We pro pose DFU-MambaKAN,alightweight hybrid architecture that combines a pure-Py Torch selective state-space (Mamba) block with a Gaussian radial-basis-function Kolmogorov Arnold Network (RBF-KAN) feed-forward layer. The model targets two clinically moti vated tasks organized as a screening-then-grading cascade: (1) binary normal-versus-ulcer screening and (2) four-class Wagner-Meggitt severity grading. It uses only 1.06 million parameters, 1.4 to 22 times fewer than four standard baselines (ResNet50, Efficient Net B0, MobileNetV3-Small, ViT-Tiny) evaluated under an identical, deduplicated, fixed-seed train/validation/test protocol. On the screening task, DFU-Mamba KAN reaches 97.17% accuracy (macro-F1 0.970, AUC 0.994), within 0.9–2.9 points of all four baselines. On the severity-grading task it reaches 67.75% accuracy (macro-F1 0.677, AUC 0.895), a much larger gap than the baselines (98.1–99.0%), which we analyze in detail and attribute mainly to under-convergence of the sequential state-space scan on a problem with ten times more data and twice as many classes, rather than to a fundamental limitation of the architec- ture. We also checked the two public Kaggle-hosted DFU datasets used here for duplicate images and found a 32.4% exact-duplicate rate in one of them; left uncorrected, this in- flates reported accuracy through train/test leakage. The main contribution of this paper is not a state-of-the-art accuracy claim. It is a reproducible, efficiency-aware benchmark, a parameter-efficient architecture, and a reported limitation, together with concrete next steps toward low-cost, edge-deployable DFU triage tools.

Abstract: The hierarchical organization of topological associating domains (TADs) is a funda-mental feature of three-dimensional genome architecture, yet its systematic character-ization remains challenging due to the scale-dependent nature of conventional detection methods. Here, we apply hyperbolic embedding into a Poincaré disc to analyze TAD hierarchies from Hi-C data. In this framework, genomic loci are projected onto the disc such that the radial coordinate encodes hierarchical depth: subTADs localize near the center, while metaTADs shift toward the periphery. The method was validated on ten cell lines, including the isogenic MCF10A/MCF7 pair, and does not require manual pa-rameter tuning across resolutions. In the MCF10A/MCF7 isogenic system, neoplastic transformation was associated with a redistribution of the hierarchy: the number of small TADs (levels 1–3) decreased by 6.7%, whereas large TADs (levels 4–6) increased by 4.7%, with the largest domains (level 6) showing a 33.9% increase. Comparison with standard approaches — Insulation Score (IS) and Directionality Index (DI) — yielded an average Jaccard index of 0.568 and F1 score of 0.718 against DI. Unlike IS and DI, which operate at fixed scales, the proposed approach recovers the full TAD hierarchy from a single embedding, enabling cross-resolution and cross-cell-line comparisons without parameter reoptimization. These results demonstrate that the Poincaré disc method provides a robust, interpretable, and scale-invariant framework for detecting hierar-chical chromatin rearrangements associated with cancer.

Abstract: Metallic wing covers are defined here as wing skins with mechanically attached or integrally machined stringers; an integrally stiffened panel is the monolithic case. This structured narrative review examines upper and lower metallic wing covers as manufacturing objects in civil transport, business-aircraft and selected military fixed-wing programmes. Public evidence, spanning shot-peening process origins in the 1920s–1930s and principal aircraft-programme evidence from the 1950s onward, was synthesised from peer-reviewed papers, SAE Technical Papers, patents, trade literature, government reports and supplier disclosures using explicit source weighting rather than statistical aggregation. Evidence was graded by source strength, and patents were treated as capability evidence rather than proof of serial production unless independently corroborated. The synthesis shows that forming-route selection is governed by structural scale, cover role, curvature class, alloy and temper, inherited stock state, panel architecture, and compensation or validation capability. Upper/lower cover divergence is directly evidenced for selected Airbus, Gulfstream and B-1B cases; where cover-separated routes are not publicly disclosed, it is treated as inference rather than production fact. The strongest public evidence supports peen and particle-impact routes for directional or inflected lower covers, creep age forming for large smooth heat-treatable covers, and specialised laser or hybrid routes where independently corroborated.

Abstract: Existing reviews on MNP removal from water rarely link adsorbent structural features to the molecular interactions governing removal performance. This review addresses this gap by examining MNP adsorption from a mechanism-oriented perspective, mapping six canonical interaction pathways across five adsorbent classes. Adsorption emerges as a system-dependent process governed by the interplay between polymer properties and surface chemistry rather than by the material alone. Interactions such as π–π stacking and hydrophobic affinity dominate for non-functionalised polymers on carbon-rich surfaces, while electrostatic forces and hydrogen bonding become more relevant for oxidised particles. Pore structure becomes significant when particle size and porosity match, whereas chemisorption provides a stronger and faster pathway in systems containing reactive metal sites. Across material classes, differences relate more closely to scalability and sustainability than to intrinsic adsorption capacity. Bio-based materials offer a favourable balance between performance and practical implementation, while more advanced systems provide greater control but remain limited by synthesis complexity. Importantly, laboratory capacities often overestimate real performance, and removal efficiency in complex matrices is a more reliable metric. Future progress will depend on improved standardisation, better integration with modelling, and validation under realistic conditions to support the transition from laboratory studies to practical applications.

Abstract: Large language models are increasingly used to read résumés and judge who advances in hiring, a task once reserved for people and now handed to systems whose reasoning is hard to inspect. Whether these models carry the demographic biases that have long shaped human hiring is therefore an urgent question, and the published evidence so far is mixed and difficult to interpret, partly because studies tend to test a single condition and rarely confirm that their measurement instrument can detect bias at all. This paper audits demographic bias in résumé evaluation across three current models, one of them open-weight, and treats robustness as a central concern rather than seeking a single verdict. The audit pairs a positive control that confirms the models read genuine differences in candidate quality with a deliberate attempt to provoke bias by weakening candidates, relaxing the prompt, and adding culture-fit language of the kind used in real hiring. Bias tied to race and gender is found to be very small and to stay small even under these adverse conditions, which is a more reassuring and more demanding result than a null measured once. The one systematic preference that emerges favours candidates who appear more experienced, and closer inspection shows that most of it is an artifact of how the résumés were built rather than a bias against age, leaving only a modest effect that surfaces when the prompt is casual. A separate and quieter pattern appears in the open-weight model, which reacts to a few explicit signals of minority status. The broader lesson is that fairness measured on a clean benchmark does not by itself guarantee fairness in deployment, because how a model is prompted can decide whether bias appears.

Abstract: The hotel industry is highly competitive, with hotels seeking new ways to increase market share and boost direct bookings against the dominance of Online Travel Agencies. While previous research has established that hotel website quality influences booking behavior, the role of website aesthetics, specifically dimensions such as photography, design, and text, has remained insufficiently theorized and tested. This paper addresses that gap by proposing and testing model which is theoretically grounded by integrating the Stimulus-Organism-Response framework, the Elaboration Likelihood Model, and Signaling Theory, to explain how website aesthetic dimensions relate to hotel features and direct booking share. The study employs a content analysis of 201 categorized Croatian hotel websites combined with Chi-square analysis and Cramer’s V effect-size estimation. Findings confirm significant associations between hotel features (rating, chain, consortia membership) and website aesthetic dimensions, and between aesthetic dimensions and direct booking share, with photography and text showing the strongest associations within the sample. These results extend theoretical understanding of website aesthetics as an antecedent of booking behavior and offer hoteliers evidence to prioritize content investment. The study also addresses methodological concerns by reporting effect sizes alongside significance levels and a structured coding protocol. Future research should explore the directionality of these relationships, identified as a limitation of this study.

Abstract: Viruses have developed a diverse array of mechanisms to hijack host transcriptional machinery and ensure successful viral gene expression. One such method is through manipulation of the cellular machinery required for post-transcriptional processing of host messenger RNAs (mRNAs). 3’ end processing of host mRNA requires a complex suite of proteins that function together to identify potential polyadenylation sites, cleave the pre-mRNA at the selected site, and synthesize the polyadenosine tail. Under certain cellular conditions - including stress, disease, and infection - altered regulation of these complexes can lead to changes in alternative polyadenylation (APA) site usage, resulting in changes in 3’ untranslated region (UTR) length, transcript abundance, and translation potential. Recent studies have identified APA as an emerging regulator of viral infection. Viruses interact with polyadenylation machinery both directly and indirectly to facilitate viral gene expression, evade innate immune responses, and achieve targeted host shutoff. While the specific interactions vary, viral manipulation of 3’ post-transcriptional processing proteins is common among a range of viruses, including herpes simplex virus (HSV), influenza A virus (IAV), and human immunodeficiency virus (HIV). In this review, we provide an overview of cellular polyadenylation machinery and mechanisms of APA that are exploited by viruses, as well as the methods that can be used to analyze changes in APA. We highlight the ways in which a range of both DNA and RNA viruses manipulate post-transcriptional processing and APA to regulate viral and host gene expression and enhance cellular permissivity to infection. Finally, we highlight current knowledge gaps and future directions exploring the translatability of these discoveries toward the development of future therapeutic strategies.

Abstract: Antimicrobial peptides (AMPs) are increasingly regarded as next-generation antimi-crobial agents because of their broad-spectrum activity, rapid killing, antibiofilm po-tential, immunomodulatory properties, and mechanisms of action that differ from those of many conventional antibiotics. Most AMPs are short, cationic, and amphipathic molecules that interact with negatively charged microbial envelopes. However, their biological activities extend beyond membrane disruption and include intracellular targeting, immune modulation, endotoxin neutralization, and interference with biofilm formation. Despite these advantages, the clinical translation of AMPs remains limited by proteolytic instability, hemolysis or cytotoxicity, poor pharmacokinetics, salt and serum sensitivity, production costs, and delivery challenges. The AMP field is now shifting from natural peptide discovery toward integrated engineering pipelines that combine rational peptide modification, biomaterial-based delivery, high-throughput screening, and artificial in-telligence/machine learning (AI/ML). Chemical and structural modifications, including D-amino acid substitution, cyclization, lipidation, PEGylation, terminal amidation, hy-drocarbon stapling, hybridization, sequence truncation, metal coordination and immo-bilization onto biomaterials, are being used to improve stability, potency, selectivity, antibiofilm activity, and tissue localization. In parallel, AI/ML approaches now enable the large-scale mining of microbiomes, extinct proteomes, venom-derived sequences, and de novo peptide sequence space. Recent predictive, generative, and optimization-based AI/ML approaches now accelerate AMP discovery by mining large biological datasets, identifying cryptic antimicrobial sequences, generating de novo peptide candidates, and optimizing multiple properties, including potency, selectivity, stability, toxicity, and synthesizability. This focused review summarizes recent advances in AMP research, highlights engineering strategies used to endow AMPs with improved biological and pharmacological properties and discusses how AI/ML is reshaping antimicrobial peptide discovery in the context of antimicrobial resistance.