Browse Articles

Comparative transcriptomic dynamics reveal molecular responses of susceptible and resistant Triticum aestivum genotypes to wheat stripe mosaic virus

The contribution of increased global soil salinity to changes in inorganic carbon

Soil salinization poses a serious environmental challenge, but the impact of global salinity on SIC (Soil Inorganic Carbon) remains unclear. Using 94,515 samples from 0 to 200 cm depth, combined with subregional classification (such as soil type, land use, climate, geomorphology, and soil texture) which helps address spatial heterogeneity, we obtain relatively accurate global distribution data for EC (Electrical Conductivity) and SIC. EC of 0 to 40 cm layer positively influences SIC in most taxonomic subregions, which may be due to the inorganic CO 2 absorption influenced by pH and salinity. EC of 80 to 100 cm layer sometimes negatively influences SIC due to the increase of soil depth. When EC is below 4 dS/m, EC often positively influences SIC. When EC increases by 2 to 4 dS/m, the mean global SIC in 0 to 20, 20 to 40, and 80 to 100 cm layers increases from 66.15, 78.75, and 117.39 to 174.47 to 190.38, 132.76 to 154.98, and 149.14 to 161.37 g/kg, respectively. The increase is relatively high but similar overall, which deserves high attention. These findings elucidate the dynamics of carbon–salt coupling in the soil–atmosphere–water system, offering pivotal scientific insights for carbon-neutrality strategies.

Diagnostic performance of multimodal biomarkers in colorectal cancer

Anharmonicity driven unusual particle-to-wave-like phonon crossover leads to ultralow thermal conductivity in Tl ₂ AgI ₃

Realization of unusual particle-to-wave-like crossover in phonon transport and understanding its fundamental structural origin can guide the design of materials with ultralow thermal conductivity. Here, we report such a crossover from particle-like phonon propagation to wave-like coherence with increasing temperature in the zero-dimensional (0D) metal halide, Tl 2 AgI 3 . Composed of discrete (Tl 6 I) 5+ and (Ag 3 I 8 ) 5− subunits, the structure exhibits intrinsic lattice instability governed by Pauling’s third rule where face-sharing of polyhedra drives Coulombic cationic repulsion causing local distortion of Ag atoms, as confirmed by synchrotron X-ray pair distribution function (X-PDF) analysis and ab initio molecular dynamics (AIMD) simulations. Anharmonic low-energy rattling of Tl is evidenced within the (Tl 6 I) 5+ framework. These structural disorders generate low-frequency localized and anharmonic optical phonons that hybridize with acoustic branches, strongly suppressing lattice thermal conductivity ( κ l ). Consequently, κ l drops to ~0.18 W/m.K at 125 K and remains nearly temperature independent, signaling a breakdown of the phonon-gas model, attributed to phonon localization and wave-like coherence, modeled using the linearized Wigner transport equation (LWTE). The phonon localization in the 0D crystal structure results in a crossover from populations conductivity ( κ p ) associated with particle-like phonon propagation to coherence conductivity ( κ c ) through wave-like tunneling, at 175 K. Our study reveals 0D structural confinement along with anharmonic local structural dynamics can enable particle-to-wave-like phonon crossover, establishing a pathway to mixed phononic regimes and suppressed thermal transport.

MQADet: a plug-and-play paradigm for enhancing open-vocabulary object detection via multimodal question answering

Abstract Open-vocabulary detection (OVD) aims to detect and classify objects from an unrestricted set of categories, including those unseen during training. Existing open-vocabulary detectors often suffer from visual-textual misalignment and long-tailed category imbalance, leading to poor performance when handling objects described by complex, long-tailed textual queries. To overcome these challenges, we propose Multimodal Question Answering Detection (MQADet), a universal plug-and-play paradigm that enhances existing open-vocabulary detectors by leveraging the cross-modal reasoning capabilities of multimodal large language models (MLLMs). MQADet can be seamlessly integrated with pre-trained object detectors without requiring additional training or fine-tuning. Specifically, we design a novel three-stage Multimodal Question Answering (MQA) pipeline that guides MLLMs to accurately localize objects described by complex textual queries while refining the focus of existing detectors toward semantically relevant regions. To evaluate our approach, we construct a comprehensive benchmark across four challenging open-vocabulary datasets and integrate three state-of-the-art detectors as baselines. Extensive experiments demonstrate that MQADet consistently improves detection accuracy, particularly for unseen and linguistically complex categories, across diverse and challenging scenarios. To support further research, we will publicly release our code.

The role of fibration symmetries in geometric deep learning

Geometric Deep Learning (GDL) unifies a broad class of machine learning techniques from the perspectives of symmetries, offering a framework for introducing problem-specific inductive biases like Graph Neural Networks (GNNs). However, the current formulation of GDL is limited to global symmetries. We propose to relax GDL to allow for local symmetries, specifically fibration symmetries, which only require isomorphic input trees—a property that is much more common in real-world graphs. We show that GNNs apply the inductive bias of fibration symmetries and derive a tighter upper bound for their expressive power. Additionally, by identifying symmetries in networks, we compress network nodes, thereby increasing their computational efficiency during both inference and training of deep neural networks. The mathematical extension introduced here applies beyond graphs to manifolds, bundles, and grids for the development of models with inductive biases induced by local symmetries that can lead to better generalization.

Characterizing surface soil heavy metal contamination and source attribution in the Qinghai Lake Basin

A shear-induced limit on bacterial surface adhesion in fluid flow

Controlling bacterial surface adhesion and subsequent biofilm formation in fluid systems is crucial for the safety and efficacy of medical and industrial processes. Here, we theoretically examine the transport of bacteria close to surfaces, isolating how the key processes of bacterial motility and fluid flow interact and alter surface adhesion. We exploit the disparity between the fluid velocity and the swimming velocity of common motile bacteria and, using a hybrid asymptotic-computational approach, we systematically derive the coarse-grained bacterial diffusivity close to surfaces as a function of swimming speed, rotational diffusivity, and shape. We calculate an analytical upper bound for the bacterial adhesion rate by considering the scenario in which bacteria adhere irreversibly to the surface on first contact. Our theory predicts that maximal adhesion occurs at intermediate flow rates: At lower flow rates, increasing flow increases surface adhesion, while at higher flow rates, adhesion is decreased by shear-induced cell reorientation.

Preparation and characterization of low-cost chemically activated carbons using H3PO4, ZnCl2 and KOH for CO2 adsorption applications

Bioresponsive immunomodulator nanocomplex for selective immunoengineering in metastatic lymph nodes

Lymph node (LN) metastasis (LNM) frequently occurs in various cancer types and is associated with high aggressiveness, poor prognosis, and low survival rates. However, effective clinical interventions remain limited primarily due to the distinctive immunosuppressive microenvironment found in metastatic LNs. Targeted delivery of immunomodulators and selective immunoengineering in metastatic LNs offers a promising avenue for repurposing these LNs as an effective antitumor system while mitigating the risk of unwanted immune activation elsewhere. Here, we develop a bioresponsive LN-targeted immunomodulator nanocomplex designed to selectively reprogram the immune microenvironment in metastatic LNs for LNM inhibition. The immunomodulator nanocomplex can target LNs due to specific chemokine receptor 7 modification and selectively release anti-PD-1 antibodies in response to glutathione that is found elevated in the extracellular matrix of metastatic LNs. In two mouse models, our data suggested that the immunomodulator nanocomplex can selectively activate T cell–mediated antitumor immune responses in metastatic LNs and thus effectively inhibit tumor growth and prolong the survival of animals. Importantly, the modular design of this platform could enable facile incorporation of alternative immunotherapeutic agents that exhibit significant systemic toxicities in the clinic, allowing broader application to payloads that may particularly benefit from localized, LNM-selective activation. This approach holds significant promises for reducing the necessity for complete LN dissection, thereby presenting a valuable therapeutic option for a broad spectrum of cancer patients.

Correction: A new twist in the evolution of chameleons uncovers an extremely specialized optic nerve morphology

David Baltimore: Scientist, leader, and mentor

Fifty years ago, at the remarkably young age of 37, David Baltimore received the Nobel Prize (with Howard Temin and Renato Dulbecco) for “discoveries concerning the interaction between tumor viruses and the genetic material of the cell.” David was a prolific scientist whose work spanned many topics, but he was first and foremost a virologist. His recent passing invites us to reflect on a remarkable intellectual trajectory that began with seminal discoveries in virology, broadened to encompass major advances in cancer biology and immunology, and culminated in a legacy—sustained by the many scientists he trained—that will continue to shape modern biomedicine for years to come.

Correction: Visualizing the dynamic polymerization of the bacterial actin-like cytoskeleton for magnetic organelle positioning

A novel augmented reality and reinforcement learning empowered communication framework for underwater unmanned autonomous vehicle

End-to-end emergency response protocol for tunnel accidents augmentation with reinforcement learning

Leveraging medical imaging and deep learning for diagnosis of breast cancer using histopathological images

Experimental study and evaluation analysis on the plugging mechanism of sand control screen in argillaceous Fine-Silt gas hydrate reservoirs

Multi-scale study on dynamic damage characteristics and energy dissipation of deep rock under thermal-hydro-mechanical coupling

Numerical study on fatigue failure mechanism of reinforced concrete slabs under coupled action of corrosion and cyclic loading

Trimester-aware yoga video recommendation using hybrid deep learning for pregnant woman