Browse Articles
Discover research articles across all indexed journals
A robust natural language text-to-SQL generation framework with dynamic strategies based on LLMs
Abstract Natural language text-to-SQL generation (Text2SQL) aims to translate natural language questions into executable SQL queries. Although the emergence of large language models (LLMs) has led to significant advancements in this field, their performance degrades sharply with question complexity increases. A key limitation of current LLM-based methods lies in their uniform generation strategies, which fail to adapt dynamically to varying question complexity. To address this issue, we propose TriSQL, a novel three-stage framework designed to analyze question complexity and generate accurate and executable SQL. First, a Question-Guided Schema Selector is conceived to get the most relevant schema to the question using cross attention. Second, a Structure-Aware SQL Generator takes both the question and the selected schema as input, employing hierarchical decoding to generate a syntactically valid initial SQL. Finally, a Complexity-Aware SQL Refiner is designed with LLM to dynamically adjust strategies corresponding to the complexity of question and initial SQL, ensuring that the final generated SQL is both accurate and executable. Experimental results on the Spider benchmark and its variants show that TriSQL achieves state-of-the-art execution accuracy, surpasses existing LLM-based methods, and provides both high efficiency and strong robustness.
MYC modulates TOP2A diffusion to promote substrate detection and activity
Abstract Topoisomerases alleviate DNA supercoiling by cleaving and resealing DNA strands. Previously, we showed that the oncoprotein MYC recruits and stimulates topoisomerases to remove DNA entanglements generated by oncogenic transcription. Understanding this mechanism may suggest methods to inhibit MYC-driven topoisomerase activation, targeting tumor-specific transcription. Here, we demonstrate that the essential topoisomerase TOP2A in human cells exists in a dynamic equilibrium between sequestration in the nucleolus, substrate searching in transcription hubs, and active engagement on chromatin. This equilibrium is highly responsive to changes in DNA topology, allowing cells to regulate TOP2A levels. Using single molecule tracking, here we show that MYC accelerates TOP2A diffusion in cells. We explain this phenotype by demonstrating that MYC limits TOP2A self-interaction in vitro, while decreasing the size of TOP2A complexes in cells. By increasing TOP2A diffusion, MYC promotes substrate binding and increases TOP2A engagement on chromatin genome-wide, revealing the mechanism underlying MYC stimulation of TOP2A activity.
Musician presence and its effects on physiological and psychological well-being in live versus livestreamed concerts
Abstract Listening to music, whether live or prerecorded, can benefit health and well-being. Studies have shown that music can alleviate pain and anxiety, improve emotional well-being, and strengthen social connectedness. While research on live music is promising, direct comparisons with recorded settings—specifically in relation to the physical co-presence of performing musicians—are limited. In this randomized controlled pilot study (n = 130), we examined the effects of live music concerts (classical and pop) compared to a simultaneous cinematic livestream, in a parallel design. Psychological and physiological well-being were assessed through self-reported questionnaires and physiological measures, including heart rate (HR) and heart rate variability (HRV). In the live condition, participants reported greater appreciation for the music, felt more moved by the music as indicated by higher kama muta scores, and reported higher arousal and valence compared to the livestream condition. Using linear mixed models, we found that only the mean HR was significantly higher in the live group, whereas other HRV parameters did not show statistically significant differences. In conclusion, the “liveness”—in terms of the musicians being physical present—emerges as an important aspect shaping audience’s subjective experiences and physiological responses, particularly HR.
The crucial but insufficient role of E2s domain’s residues 490 and 492 in determining the host tropism of hepatitis E virus
Abstract Hepatitis E virus (HEV) is a significant pathogen causing acute viral hepatitis globally, posing a particular threat to pregnant women. HEV infects a range of host species, with distinct genotypes exhibiting genotype-specific tropism for different host hepatocytes. The P domain of viral capsid protein plays a central role in host cell attachment, but the molecular determinants that govern its host specificity remain unclear. This study investigates the molecular mechanisms underlying HEV host tropism by using a zoonotic HEV specific antibody 6H8. An epitope involving residues 490 and 492 is identified crucial for both mAb 6H8 binding and virus-cell attachment. Structure-based mutagenesis, molecular dynamics simulations, virus-cell attachment assays, and viral infectivity assays highlight the importance of the N490 and M492 residues in maintaining the structural integrity of the 6H8 epitope, influencing host specificity. Mutations at 490 and 492 permit HEV-1’s and disrupt HEV-4’s binding and infection in porcine hepatocytes. However, they are insufficient alone for reestablishing swine infection in vivo, indicating additional factors are involved in HEV’s host tropism. Our findings suggest that N490 and M492 are critical but not sole determinants of HEV-4’s specific tropism for porcine hepatocytes and advance our understanding of HEV’s zoonotic transmission and host specificity.
Experimental investigation on solar air heating system using evacuated tube collector with coaxial tube
Photocatalytic Stereoselective Editing of Alkynes to 3D Molecules via Hydrogen Atom Transfer-Mediated Dynamic Epimerization
Miniaturized RF reconfigurable bandpass filter with dynamic wideband frequency and constant bandwidth tuning capability
Devaluation of response-produced safety signals reveals circuits for goal-directed versus habitual avoidance in dorsal striatum
EEG-based predictors of motor recovery during immersive VR-BCI rehabilitation
Ultrafast solvent-modulated roaming mechanism in bromoform revealed by femtosecond X-ray solution scattering
Integrated virtual screening, ADMET profiling, and molecular dynamics simulations of novel natural HDAC6 inhibitors with the potential to ameliorate skeletal muscle degeneration
Vaccination against H5 HP avian influenza virus leads to persistent immune response in wild king penguins
Abstract Since 2021, the panzootic nature of high pathogenicity avian influenza (HPAI) represents an increasing threat to wild vertebrate populations. In this context, recent vaccines developed for poultry could provide tools for the conservation of wild endangered birds populations. The king penguin ( Aptenodytes patagonicus ), a long-lived seabird breeding in dense colonies with an extended chick-rearing period, has been identified as a possible surrogate species for a vaccination trial in a sub-Antarctic natural setting. Here we investigate the immune response of king penguin chicks to a self-amplifying mRNA vaccine against a H5 HPAI clade 2.3.4.4b protein. The cohort entails thirty vaccinated chicks (primo- and boost-injections), and 20 unvaccinated controls. Along 250 days of monitoring, the vaccinated chicks show a high and persistent immune response, granting a strong sero-neutralisation capacity against the virus, up to fledging. No adverse effects are observed. Screening for antibodies against unspecific avian influenza viruses suggests that no natural infection has occurred over the entire trial. The emergence of HPAI in the Southern Indian Ocean in October 2024 highlights the timeliness of such experimental tests. Our results thus show the vaccine could provide a potentially powerful tool for mitigation of avian flu outbreaks in the wild.
The astrocyte marker ALDH1L1 also identifies a stromal cell population in the lymph node
Abstract ALDH1L1 is widely used as a marker of astrocytes in the central nervous system (CNS), but its expression and potential roles in the periphery, particularly in lymphoid organs, remain poorly understood. Here, we found that ALDH1L1 + cells comprise approximately 5–9% of the lymph node (LN). To better understand their identity, we investigated whether these cells share characteristics with glial, immune, endothelial or stromal cell populations under homeostatic conditions. Using ALDH1L1/TdTomato reporter mice, immunofluorescence, and flow cytometry, we found that TdTomato + cells in LNs do not coexpress canonical CNS (GFAP, ACSA-2) or peripheral glial (Sox10) markers. Similarly, using multiple approaches we found minimal overlap with T cells (CD3, TCRβ), B cells and dendritic cells (B220), myeloid cells (CD11b, Iba1), or antigen presenting cells (MHCII). To explore potential stromal associations, we assessed coexpression of TdTomato with endothelial (CD31, LYVE1) and reticular (ER-TR7) markers and found insubstantial overlap. In contrast, TdTomato + cells showed significant colocalization with podoplanin (PDPN), a marker of fibroblastic reticular cells (FRCs). Interestingly, while many TdTomato + cells were PDPN + , relatively few PDPN + cells expressed TdTomato. Notably, ALDH1L1 + PDPN + cells were enriched in the paracortex and medulla but absent from the B cell follicles and subcapsular sinus. These findings suggest that while ALDH1L1 remains a reliable marker of astrocytes in the CNS, it may also delineate a distinct subset of FRCs in the LN, opening new avenues for exploring stromal cell diversity and function in lymphoid tissues.
Using firm-level supply chain networks to measure the speed of the energy transition
Abstract International climate targets rely on the success of the energy transition, however systematic monitoring on how the economy adopts low-carbon energy remains underdeveloped. Here we use nationwide supply-chain network data to reconstruct energy portfolios for 25,000 Hungarian firms between 2020 and 2024, covering 75% of gas, 70% of electricity, and 50% of oil consumption. This allows us to quantify the speed of the energy transition -the transition towards low-carbon electricity- on the firm level. We find substantial heterogeneity in decarbonization progress: half of firms increase low-carbon energy shares, but 50% reduce it. Energy cost structures are closely associated with transition behavior, indicating technology-related lock-in effects. Extrapolating current trends yields an aggregate low-carbon share of 20% by 2050, highlighting ineffective decarbonization efforts. If firms strictly adopted strategies of decarbonization frontrunners within their industry sectors, a low-carbon share of 70% could be achieved by 2050, putting climate targets within reach.
Simulated depression risk classification from Parkinson’s voice features using a self-attention-enhanced MLP architecture
Abstract Parkinson’s disease affects both motor and non-motor functions, including vocal features that may indicate underlying mental health conditions such as depression. This work proposes a novel framework for simulated depression risk classification using vocal biomarkers derived from the UCI Parkinson’s dataset. A Self-Attention-Enhanced Multilayer Perceptron-MLP architecture is used model interactions between key acoustic features, particularly Harmonic-to-Noise Ratio and Jitter, which serve as the basis for generating binary depression risk labels. The proposed model outperforming traditional and deep learning benchmarks including Support Vector Machine (SVM), k-Nearest Neighbors (k-NN), TabNet, CNN-LSTM, Deep Neural Network (DNN), and Explainable Boosting Machine (EBM) with an accuracy of 97%, F1-score of 98%, recall of 95%, and specificity of 100%, While EBM offers strong interpretability, the attention-enhanced model demonstrates optimal predictive capability. These findings highlight the efficacy of voice-based features combined with attention mechanisms for early, non-invasive identification of depression risk in PD patients.
A human electrophysiological signature of Fragile X pathophysiology is shared in V1 of Fmr1-/y mice
Abstract Predicting clinical therapeutic outcomes from animal studies using conserved electrophysiological phenotypes could facilitate developing treatments for neuropsychiatric disorders. Alpha oscillations in human resting-state electroencephalogram recordings are altered in many disorders, but whether these disruptions exist in mouse models is unknown. Here, we employed a uniform analytical method to show in males with fragile X syndrome (FXS) that alpha oscillations in humans and alpha-like oscillations in the visual cortex of Fmr1 -/y mice are slowed, with a stronger phenotype in adults than juveniles and a juvenile-specific power phenotype in both species. We find that alpha-like oscillations are disrupted by deletion of Fmr1 in cortical excitatory neurons and glia, reflect differential activity of two classes of GABAergic interneurons, and are more sensitive to activation of GABA B receptors by Arbaclofen in wild-type than Fmr1 -/y mice. Our framework reveals evolutionary conservation of alpha oscillation disruptions, enables a deeper understanding of FXS pathophysiology, and narrows the gap between treatment promise and practice.