Browse Articles

DxDirector: an agentic large language model driving the full-process clinical diagnosis

Abstract Clinical diagnosis in the real world often begins with ambiguous patient complaints that require iterative reasoning and testing. While large language models (LLMs) increasingly assist with specific medical queries, they currently lack the ability to autonomously drive this entire diagnostic workflow, limiting their potential to significantly alleviate physician workload. Here we present DxDirector-7B, an agentic LLM designed to navigate the full diagnostic process through advanced slow thinking capabilities. Unlike existing assistants, our model autonomously determines optimal diagnostic strategies, requesting physician intervention only for necessary clinical operations. In evaluations spanning rare diseases and complex real-world cases, DxDirector-7B achieves superior diagnostic accuracy compared to state-of-the-art medical and general-purpose LLMs with significantly larger parameters. Crucially, it drastically reduces physician involvement while maintaining a robust safety and accountability framework for high-risk conditions. These results demonstrate a paradigm shift where AI effectively leads clinical reasoning, offering a scalable solution to enhance diagnostic efficiency and accessibility.

Neoadjuvant toripalimab plus CapeOX in locally advanced Epstein-Barr virus-associated gastric or gastroesophageal junction adenocarcinoma: a phase Ⅱ trial

Electrochemical corrosion accompanies dendrite growth in solid electrolytes

The structure of the Vibrio alginolyticus flagellar filament suggests molecular mechanism for the rotation of sheathed flagella

Abstract In several pathogenic bacteria, including Vibrio species, the filament of the bacterial flagellum is encased by a membranous sheath, an extension of the bacterial outer membrane. It has been proposed that having sheathed flagella permit bacteria to evade an immune response against flagellar components, suggesting a role in virulence. However, the molecular details of the interaction between sheath and filament, and how it impacts filament rotation, remain largely uncharacterized. Here, we combine single-particle cryo-electron microscopy, cryo-electron tomography, and genetic analyses to resolve the molecular architecture and biogenesis of the sheathed flagellum in Vibrio alginolyticus . We show that the flagellar filament forms a canonical 11-stranded supercoil made of the flagellin FlaD2 and enveloped by a bilayered sheath. We report that the filament surface is highly electronegative, suggesting that electrostatic repulsion between filament and sheath may reduce friction and supports high-speed flagellar rotation. We also show that the filament cap protein FliD possesses a unique domain in sheathed flagella, that may coordinate sheath assembly with filament elongation. Collectively, this structural insight into the structure of the Vibrio alginolyticus flagellum suggests a molecular mechanism for the rotation of sheathed flagella.

Machine learning-driven alignment architecture of heterogeneous data with transient varying semantics

Abstract Via cross-correlation algorithms or synchronized acquisition of signals, the alignment of heterogeneous data with unknown semantic time shifts and intermittent semantic variations cannot be solved. The shift is caused by different data acquisition principles of sensors, different response discrimination principles using heterogeneous data, etc. Here, we report an unsupervised alignment architecture with a supervised learning model as the kernel to overcome the limitations of brain cognition, perception, and storage in aligning complex heterogeneous data. A set of data with a time shift is input into the kernel model of the architecture to predict the semantic labels, features or continuous values corresponding to another set of data. The time shift corresponding to the maximum testing accuracy or the minimum mean squared error is the alignment parameter for the two heterogeneous datasets. This architecture is expected to serve as a preprocessing step for semantic mining of signals and for information fusion.

Dominant clones leverage developmental epigenomic states to drive ependymoma

In silico discovery of nanobody binders to a G-protein coupled receptor using AlphaFold-Multimer

Abstract Antibodies are central mediators of the adaptive immune response, and they are powerful research tools and therapeutics. Antibody discovery requires substantial experimental effort, such as immunization campaigns or in vitro library screening. Predicting antibody-antigen binding a priori remains challenging. However, recent machine learning methods raise the possibility of in silico antibody discovery, bypassing or reducing initial experimental bottlenecks. Here, we report a virtual screen using AlphaFold-Multimer (AF-M) that prospectively identified nanobody binders to MRGPRX2, a G protein-coupled receptor (GPCR) and therapeutic target for the treatment of pseudoallergic inflammation and itch. Using previously reported nanobody-GPCR structures, we identified a set of AF-M outputs that effectively discriminate between interacting and non-interacting nanobody-GPCR pairs. We used these outputs to perform a prospective in silico screen, identified nanobodies that bind MRGPRX2 with high affinity, and confirmed activity in signaling and functional cellular assays. Our results provide a proof of concept for fully computational antibody discovery pipelines that can circumvent laboratory experiments.

The network structure of cross-feeding impacts microbial community diversity under growth-inhibiting stresses

Necking of the active Turkana Rift Zone and the priming of eastern Africa for continental breakup

Multiomics immune profiling of a patient-relevant orthotopic lung cancer model using SEPARATE-Seq

High-throughput single-vesicle imaging platform for direct extracellular vesicle profiling of human plasma

Type I interferon drives T cell responses to amyloid beta in the central nervous system

Abstract Amyloid beta (Aβ) plaque deposition in the central nervous system (CNS) is a hallmark of Alzheimer’s disease (AD) and cerebral amyloid angiopathy (CAA), triggering robust innate immune responses. However, the role of the adaptive immune system remains less well understood. Here we show the immune microenvironment dynamics in APP23 transgenic (APP23-tg) mice modelling CNS amyloid pathology, using single-cell transcriptomics. We observed a marked increase in T-cell populations during late disease stages, particularly CD8⁺ T-cells that clustered around Aβ plaques, suggesting a targeted immune response. Among these, we identified an Aβ plaque-associated subset of CD8⁺ T cells expressing interferon-stimulated genes (ISGs), which promoted Type-I interferon signaling. This subset also produced CXCL10, facilitating the recruitment of non-ISG T cells through the CXCL10-CXCR3 axis. Importantly, similar Type-I interferon responses were detected near plaques in human CNS amyloid pathology. Together, these findings highlight a shift from microglia-driven to T-cell-mediated neuroinflammation as amyloid pathology progresses, with implications for time-resolved therapy development.

Phosphatidylserine-everted erythrocyte membrane vesicles enhance efferocytosis and remodeling of vascular grafts

Small RNA genomics of Aedes aegypti mosquitoes discovers infectious viruses that trigger an RNA interference response

Applying asymmetric-waveform alternating current in nickel-catalyzed asymmetric reductive cross-coupling

EBF2 condensates enable adipose thermogenesis through ZFP423 sequestration and epigenetic remodeling

Mapping glioblastoma’s isoform diversity using long-read single-cell analysis

CAR-CD34 (+) hematopoietic stem/progenitor cells produced in vivo protect against thoracic aortic aneurysm and dissection

Real-time nanomolar vitamin monitoring in sweat using an electrochemical skin-attached device

Single cell transcriptional evolution of myeloid leukemia of Down syndrome

Abstract Children with Down syndrome have a 150-fold increased risk of developing myeloid leukaemia (ML-DS). Unusually for a childhood leukaemia, ML-DS arises from a preleukaemic state, termed transient abnormal myelopoiesis (TAM), via a conserved sequence of mutations. Here, we examine the relationship between the genetic and transcriptional evolution of ML-DS from natural variation; a rich collection of primary patient samples and foetal tissues with a range of constitutional karyotypes. We distil transcriptional consequences of each genetic step in ML-DS evolution, utilising single-cell mRNA sequencing, complemented by phylogenetic analyses in progressive disease. We find that transcriptional changes induced by the TAM-defining GATA1 mutations are retained in, and account for most of the ML-DS transcriptome. The GATA1 transcriptome pervades all stages of ML-DS, including progressive disease that had undergone genetic evolution. Our approach delineates the transcriptional evolution of ML-DS and provides an analytical blueprint for distiling consequences of mutations within their pathophysiological context.