Substations are critical infrastructures for ensuring reliable power system operation. With increasing digitalization and system complexity, the rapid growth of multi-source data poses significant challenges for accurate and timely fault diagnosis. Existing approaches often struggle to effectively integrate heterogeneous data or adapt to varying operating conditions. To address these limitations, this study proposes a digital twin-driven fault diagnosis framework incorporating a multi-modal fusion model that integrates system topology, alarms, fault waveforms, and SCADA data through Graph Attention Networks and self-attention mechanisms. In this work, the method is validated on a 110 kV substation using 11,597 training and 3,890 testing scenarios generated in CloudPSS. Experimental results demonstrate over 95% accuracy in fault location, 97% in fault type identification, and 90% accuracy in protection failure detection under 30% data loss conditions. The deployed digital twin system further verifies the practical feasibility of the proposed approach, highlighting its robustness in complex operating environments.

Marine heatwaves can interact across ocean basins through atmospheric teleconnections, but such inter-basin links remain poorly understood. Here we show that boreal spring and summer marine heatwaves in the Caribbean Sea are significantly connected to marine heatwaves in the Indian Ocean during the preceding winter. Using observational data and climate model simulations, we identify that boreal winter Indian Ocean marine heatwaves trigger atmospheric convection, generating an eastward-propagating Indo-Pacific-Atlantic Rossby wave train. This wave train alters the regional Hadley circulation and renders the overlying atmosphere warm, moist, and stable over the Caribbean Sea, reducing latent heat loss from the ocean, driving sea surface warming and promoting the intensification of marine heatwaves. These findings reveal a robust teleconnection pathway that allows climate signals from one basin to amplify marine heatwaves in another.

Chemical warfare agents (CWAs) pose severe global threats due to their historical role in mass casualties and ongoing misuse, particularly blister, nerve, and blood agents. To address the urgent need for field-deployable detection tools, we develop TrioAlert, which, to the best of our knowledge, is the first fluorescent sensor capable of simultaneously detecting the blister agent sulfur mustard, the G-type nerve agent sarin, and the blood agent cyanide. The sensor employs an imidazole moiety for the recognition of sulfur mustard and sarin, while cyanide is detected through a reaction at the 4-position of the coumarin fluorophore. TrioAlert demonstrates a rapid response (within 90 s), high sensitivity (limits of detection: 58.5 nM for sulfur mustard, 46.2 nM for sarin, and 53.4 nM for cyanide), and distinct fluorescence and color changes. The detection performance of TrioAlert for the three chemical warfare agents remains stable in natural environmental matrices and across various temperatures. Furthermore, TrioAlert-loaded test strips enable smartphone-based qualitative and quantitative on-site analysis, with visual detection limits of 0.05 ppm for sulfur mustard, 0.01 ppm for sarin, and 5 ppm for cyanide. The sensor also enables the visualization of all three CWAs in live cells and mouse models. Notably, in vivo fluorescence changes precede pathological damage, with distinct fluorescence triggered at sub-pathological doses. This provides an early visual indication for trace CWA exposure. Furthermore, these fluorescence changes correlate with subsequent alterations in pathological biomarkers (P < 0.01), confirming the sensor's potential for clinical diagnosis. Our study demonstrates the on-site rapid quantitative detection of high-risk chemical warfare agents, thereby advancing the capability to prevent chemical threats and safeguard public health and security.

Phosphorylation is a central post-translational mechanism for on-demand regulation of protein function. In the ATP-dependent molecular chaperone Hsp90, multiple phosphorylation sites have been implicated in activity control, yet how individual sites encode regulatory instructions remains unclear. Here, using solution NMR spectroscopy, we delineate how site-specific phosphorylation distinctly reshapes the conformational energy landscape of Hsp90 across its ATPase cycle. The phospho-mimetic mutation T115E induces a global redistribution of the N-terminal domain energy landscape, flattening conformational barriers, pre-populating a lid-closed-like excited state in the apo form, weakening ATP-driven stabilization, and impairing ADP-mediated resetting. In contrast, T36E preserves the overall structure but selectively rewires dynamics at the ATP-bound step, where accelerated exchange and a reduced excited-state population bias the ensemble toward the ground state. Together, these findings reveal that phosphorylation sites engage different dynamic allosteric routes, yet converge on a common functional outcome by suppressing productive progression through the Hsp90 ATPase cycle.

Type 2 diabetes is rising globally, with sub-Saharan Africa facing the steepest projected increase. In low-resource settings, limited screening contributes to high rates of undiagnosed disease, delaying intervention. Traditional risk models often rely on single-factor thresholds or population averages, which can miss combinations of demographic, anthropometric, and lifestyle factors relevant in African contexts. However, emerging evidence suggests that interactions between multiple risk factors provide more accurate, region-specific characterizations of risk. We applied a multidimensional subgroup discovery algorithm to cross-sectional data from three African populations and identified combinations of risk factor cutoffs that define high-risk subgroups. The most consistent profile, such as waist-to-hip ratio greater than 0.9, physical activity less than or equal to 2448 metabolic equivalent-minutes per week, and family history, shows elevated risk across regions and outperforms guideline-based definitions. Here we show that core anthropometric and familial factors are stable across populations, while others are region-specific, improving predictive performance for screening.

SARS-CoV-2 is now endemic, with infections commonplace. While much of the population now has immunity, some subsets remain at risk. For such individuals, small-molecule antivirals are the frontline treatment. However, studies have identified resistance-conferring mutations to these compounds, and there are now cases of resistant viruses emerging during treatment. These occurrences make clear the need to understand the resistance mechanisms for SARS-CoV-2 antivirals. Here, we report the pathways to resistance for atilotrelvir and simnotrelvir, two 3CL protease inhibitors used for COVID-19 treatment, and ibuzatrelvir, a compound in late-stage clinical development. Through high-throughput passaging, we reveal that resistance can readily arise, and that there is a large degree of overlap in the mutations which emerge. Moreover, viral inhibition assays demonstrate that there is not only strong cross-resistance between the emerged viruses against these three molecules, but also against two additional widely used antivirals, nirmatrelvir and ensitrelvir, as well. Cellular assays highlight S144A, E166A, and E166V as mediating broad resistance, with E166V having the strongest effects. These results have important clinical implications, including the need to carefully consider cross-resistance properties in salvage therapy and combination treatment, as well as emphasizing the need for the further development of SARS-CoV-2 antivirals with differing modalities.

Microporous framework materials, particularly in their nanosheet form, are promising building blocks for the construction of continuous molecular-sieving membranes and films, enabling a wide range of applications. However, the limited availability of such microporous nanosheets has restricted their broader use in the construction of microporous membranes. Herein, we introduce a topotactic conversion and induction (TCI) strategy that enables the rapid fabrication of ultrathin, continuous zeolite membranes on commercial substrates from heterogeneous zeolite nanosheets, thereby broadening the applicability of nanosheets in gas separation membranes. The versatility of the TCI route is demonstrated by constructing continuous CHA and MEL zeolite membranes through MFI-to-CHA and MFI-to-MEL conversions. This TCI route has the potential to enhance the fabrication of ultrathin, continuous and nanosheet-based microporous membranes and films for diverse applications. Using this route in an active gel, the all-silica CHA (Si-CHA) membrane exhibits CO2 permeance of ~8800 GPU and CO2/CH4 selectivity exceeding 1000. This separation performance index is substantially higher than previously reported values for CO2/CH4 separation. Additionally, the fabricated membranes demonstrate high separation performance among reported membraneconfigurations for CO2/N2 separation. Techno-economic analysis indicates that these membranes could reduce CH4 decarbonization cost by up to ~87% relative to amine absorption.

The environmental impact of non-recyclable thermoset waste demands sustainable polymer solutions. Existing dynamic covalent boronic ester chemistries operate at relatively low temperatures (<150 °C), resulting in insufficient stability to meet thermoset performance requirements. Here, we overcome these limitations through a N-iminodiacetic acid (N-IDA) boronate system featuring a strengthened internal B-N coordination bond. This enhanced B-N bond (~1.65 Å) endows the dynamic covalent bonds with both exceptional thermal stability and controllable exchange above 150 °C. We leverage this chemistry to develop radical crosslinkers compatible with commercial monomers and thermoplastics. The resulting recyclable thermosets demonstrate exceptional thermal stability, chemical durability under harsh environments (85 °C/85% RH, 7 days), and excellent dimensional stability. This stability-exchange dichotomy is exemplified in heat-shrinkable crosslinked polyethylene (XLPE), where the enhanced B-N coordination maintains network integrity at service temperatures (<150 °C) while enabling thermal reprocessability above 150 °C. By extending the operational range of boronic ester-based dynamic chemistry to high temperature domains, this work provides a scalable pathway toward recyclable thermosets without stability compromises.

Functional DNA nanoassemblies (fDns) hold promise for biomedical and translational applications, but broader deployment is limited by manufacturing barriers that constrain scalability, reproducibility, and stability. Current biosynthetic approaches rely heavily on downstream enzymatic processing and purification, which dominate economic, operational, and environmental costs at scale. Here we report an integrated manufacturing strategy that combines programmed circular single-stranded DNA with bacteriophage-driven biosynthesis to enable direct production of high-quality fDns with minimal post-assembly purification. By reducing reliance on purification as a major bottleneck, this approach delivers topology-enhanced stability, high assembly efficiency, and batch reproducibility, achieving sub-gram laboratory production at ~$6 per milligram while suggesting a pathway toward scalable manufacturing. We demonstrate versatility through 11 distinct fDns, including an aptamer-functionalized construct that neutralizes SARS-CoV-2 pseudovirus and authentic Omicron variants in cellular and animal models. This work provides a cost-effective, scalable manufacturing framework for fDns, advancing their sustainable and reproducible deployment across life-science applications.

Mesophase pitch-based carbon fibers (CFs) with ultrahigh axial thermal conductivity can significantly enhance through-plane heat transfer in thermal interface materials by vertical alignment, yet raise short-circuiting and electromagnetic interference (EMI) risks. Although incorporating insulating fillers or coating CFs can suppress electron migration, it often compromises thermal performance. Herein, we engineer multi-functionally trunk-branch hierarchical heterostructures to address above limitations. Vertically aligned trunk-like CF scaffolds ensure superior through-plane heat transfer. Branch-like boron nitride (BN) networks optimize impedance matching by enhancing electrical insulation and consequently improve electromagnetic wave (EMW) absorption. Meanwhile, in situ BN networks interconnect CF scaffolds, extending bidirectional thermally conductive and EMW propagation paths. At only 20.17 vol% filler, the composite exhibits a through-plane thermal conductivity of 57.96 W·m-1·K-1 (specific thermal conductivity enhancement of 1431.62 %·(vol%)-1) and an in-plane thermal conductivity of 2.93 W·m-1·K-1, together with excellent electrical insulation and absorption-dominated EMI shielding, addressing the challenge of thermal-electrical-electromagnetic coupling for next-generation electronics.