Biology subjects

Zhao, Z.

Publications and source records attributed to Zhao, Z..

At least 73 records · Page 4Linked to original sources

bioRxiv2026

ClumPyCells resolves spatial aggregation in complex tissues overcoming size biases

The spatial arrangement of cells within a tissue microenvironment shapes their interactions and cell states, which are essential for tissue development, homeostasis, and disease. Spatial -omics technologies can precisely map the location of each cell within complex tissue structures, while also profiling their protein content and transcriptional diversity. Various approaches have been developed to analyze spatial patterns of cell aggregation, repulsion, or random distribution within tissues. However, differences in cell morphology within a tissue can introduce significant bias. Cell size in particular is not accounted for and introduces challenges when quantifying the aggregation of cells or their molecular features. To overcome such limitations, we present ClumPyCells: a statistical framework that measures cell and marker aggregation within tissue while correcting for size morphology. ClumPyCells enables interpretation of cell aggregation, bypassing interfering cell types or tissue regions unrelated to the desired spatial correlation. We demonstrate the capabilities of ClumPyCells across several tumor types, including melanoma and colorectal cancer, and spatial -omics technologies such as spatial transcriptomics and proteomics, while benchmarking how cell-size differences contribute to misinterpretations. By correcting for disruptive cell types within a region of interest, ClumPyCells will determine new tissue patterns and structures without morphological interference.

bioinformatics↗

bioRxiv2026

A Pangenome Centralized NLRome Drives Lineage Specific Diversification and Functional Differentiation in Solanoideae

The Solanoideae subfamily comprises numerous economically important crops that deploy a diverse repertoire of nucleotide-binding leucine-rich repeat receptors (NLRs) to defend against pathogens. However, the evolutionary trajectory of these NLR immune receptors remains poorly understood. Here, we dissect NLR family evolution across 23 Solanoideae species by integrating genomic and transcriptomic data. We uncover a distinctive pangenome-centralized architecture wherein a mere 11.56% of soft-core orthogroups contribute 67.81% of the total NLR repertoire, with copy-number variation driven by lineage-specific expansions rather than stochastic distribution or genome size. This finding indicates that Solanoideae plants counteract biotic stress through targeted amplification of specific key families rather than undergoing broad diversification. Phylogenomic analysis resolves NLRs into four subclasses, revealing CC-NLR dominance across all species alongside extreme TIR-NLR degeneration in certain lineages, suggesting profound immune adaptive shifts. Mechanistically, recent NLR expansions are primarily fueled by proximal and tandem duplications, whereas whole-genome duplications contribute in a ploidy- and time-dependent manner, with WGD-derived genes undergoing extensive loss during diploidization. Transcriptional profiling further demonstrates duplication-type-dependent expression patterns, enabling a robust, multilayered defense network upon pathogen infection. Collectively, our study delineates the asymmetric evolutionary trajectory of the Solanoideae NLRome and establishes a framework model that provides new insights into the adaptive evolution of plant immune systems and resistance gene discovery.

genomics↗

bioRxiv2026

A Dynamic Oligomerization Network Coordinates Hemagglutinin-Mediated Membrane Fusion on Influenza Virions

Influenza A virus (IAV) entry is mediated by the trimeric glycoprotein hemagglutinin (HA), which undergoes low-pH-triggered conformational rearrangements to drive membrane fusion. Although biochemical and biophysical studies have long suggested that HA trimers function cooperatively, the structural basis for this cooperativity on intact virions has remained unclear. Here, using cryo-electron tomography (cryo-ET) and subtomogram averaging (STA), we determine the in situ structure of prefusion HA on native virions at 3.58 [A] resolution, enabling atomic modeling of membrane-anchored HA. We show that neighbouring HA trimers engage in lateral interactions mediated by HA1 subunits, assembling into flexible dimeric, pentameric, and hexameric organizations that form locally ordered lattices on the viral surface. A 6.0 [A] reconstruction of the HA-dimer reveals the molecular interfaces underlying these assemblies. Disruption of the dominant HA1-HA1 interaction markedly impairs viral entry and slows membrane fusion kinetics. Together, these findings define a virion-level mechanism for coordinated HA activation and establish a structural framework for understanding cooperative membrane fusion by class-I viral fusion proteins.

biophysics↗

bioRxiv2026

A cerebellar cognitive rheostat bidirectionally controls attention

Attention requires filtering distractors and amplifying signals, processes classically attributed to cortico-thalamic networks. Here, we reveal that the cerebellum operates as a bidirectional "cognitive rheostat" to optimize attentional states. In mice, the anterior and posterior cerebellar vermis exert opposing control over attention. Granule cells in the anterior vermis are functionally suppressed to gate sensorimotor noise via reticular nucleus-driven feedforward inhibition. Conversely, posterior granule cells are recruited by pontine inputs to amplify cognitive signals, a process relying on Grin1-mediated NMDA receptor plasticity. Circuit-specific manipulations targeting this push-pull mechanism, or localized pharmacological modulation, successfully rescued attentional deficits in an ADHD mouse model. These findings fundamentally expand the cerebellums role beyond motor coordination, identifying a topographic circuit algorithm essential for cognitive control.

animal behavior and cognition↗

bioRxiv2026

Genome-scale functional mapping of the mammalian whole brain with in vivo Perturb-seq

Functional genomics studies have provided critical insights into cell type-specific gene regulatory programs, but to date most have been conducted in wild-type tissues or cell cultures. Here, we present a gene expression functional atlas across the mouse brain. We use an enhanced in vivo Perturb-seq platform to analyze transcriptome-wide responses to loss of 1,947 disease-associated genes, profiling over 7.7 million cells spanning major brain regions and neuronal populations. We find striking cell-type-specific essentiality and transcriptional programs and show that closely related disease genes such as two NMDA receptor subunits can drive opposing transcriptional programs. Together, this work reveals insights into the genetics and mechanisms of neurodevelopmental, psychiatric, and neurodegenerative diseases in vivo, paving the way for the design of future genetic medicine.

genomics↗

bioRxiv2026

Loss of C3 and CD14 reduces region-specific neuroinflammation in a murine polytrauma model

BackgroundTraumatic brain injury (TBI) together with non-cerebral injuries characterizes the TBI-polytrauma (P-TBI) constellation, which is associated with acute neurological deterioration, delirium and unfavourable prognosis. It is hypothesized that systemic inflammatory mediators my enhances the focal, cerebral neuroimmune reaction with overall detrimental consequences, in particular in terms of acute microglial reactivity. MethodsWe explored the role of the Complement factor 3 (C3) and of the TLR-co receptor cluster of differentiation (CD14) in a murine polytrauma model that involves a mild TBI together with femur fracture, blunt thorax trauma and resuscitated haemorrhagic shock, making use of mice genetically lacking either C3, CD14 or both. ResultsWe show that P-TBI results in a rapid (4h) and brain-wide induction of inflammatory cytokines, although with distinct profiles (TNF and CCL2 having brain-wide involvement and IL-1{beta} restricted to ipsilateral cortex and striatum). TNF and CCL2 mRNA as well as protein synthesis were upregulated in microglia upon P-TBI in cortex, hippocampus and striatum which was fully abolished in the C3-/-CD14-/-animals. The analysis of single-KO animals revealed that induction of TNF and CCL2 was prevented in animals lacking C3, but not CD14, in the contralateral cortex and striatum, with an abolishment in hippocampus in mice lacking both C3 and CD14. In the cortical area of focal lesion neither C3 nor CD14 affected the induction of pro-inflammatory cytokines. ConclusionThus, C3 and CD14 are dispensable for the acute cytokine response to P-TBI in the site of injury but play differential roles across the cortex, hippocampus and striatum for the induction of cytokines in the non-injured parenchyma and in particular in microglia. Thus, interventions on C3 (mainly) and/or CD14 may reduce the encephalopathy risk associated with P-TBI but not the acute response in the injury site, where additional DAMP signalling may offer redundant activation pathways.

neuroscience↗

bioRxiv2026

In vivo multiomic Perturb-seq with enhanced nuclear gRNA capture

In vivo CRISPR screening with joint transcriptomic and chromatin readouts has been limited by inefficient recovery of gRNAs from nuclei. Here, we developed in vivo multiomic Perturb-seq, an effective platform combining nuclear transcript anchoring with gRNA-specific capture and amplification to enable high-fidelity, high-recovery gRNA assignment and scalable perturbation-resolved single-nucleus multiomics. Applying this platform to interrogate neurodevelopmental disorder risk genes in the developing cortex reveals cell-type-specific transcriptomic and epigenomic perturbation phenotypes.

genomics↗

bioRxiv2026

DCS Tools: A high-performance, resource-efficient and scalable computing suite for population-scale genomic analysis and data compression

The advent of population-scale genomics has created significant bottlenecks in computational infrastructure and storage. Traditional BWA-GATK Best Practices require massive resources, while most existing acceleration solutions depend on specialized hardware like GPUs or FPGAs, increasing costs and limiting deployment. To address this, we developed DCS Tools, a high-performance suite optimized for standard CPU architectures. DCS Tools processes a 30X WGS sample from raw FASTQ to variant calls in just 1.79 hours on a 32-thread instance--a 16-fold speedup over traditional pipelines without additional hardware. The suite integrates DPGT for million-scale joint calling and introduces SeqArc and VarArc, which reduce storage footprints by up to 80% (FASTQ) and 66% (VCF) compared to GZIP. DCS Tools offers a cost-effective, hardware-agnostic solution for petabyte-scale genomic analysis.

bioinformatics↗

bioRxiv2026

Heat stress-induced condensation of G3BP1 in perinuclear P-bodies in C. elegans' germline

Stress granules (SGs) are essential subcellular assemblies that enable cells to adapt to environmental stress, and their dynamic assembly and disassembly are critical for maintaining cellular homeostasis and reproductive capacity. However, the function and regulation of stress granules in germline cells remain mysterious. In this study, we characterized the function of GTBP-1, the homolog of human G3BP1, in response to heat stress in Caenorhabditis elegans and elucidated its regulation in stress granule dynamics. gtbp-1 mutants exhibit pronounced temperature-dependent biodirectional reproductive characteristics. At lower temperatures, their brood size is higher than that of wild-type animals, whereas at 25 {degrees}C they are completely sterile. GTBP-1 is diffusely distributed at normal culturing temperatures but forms perinuclear stress granules upon heat shock in the germline. While the NTF2 domain is essential for germline stress granule formation, other domains, including the IDR, RRM and RGG, are required for the recovery phase after heat shock. GTBP-1 stress granules colocalize with the P-bodies, but not with other germ granules. The depletion of P-bodies prohibited the perinuclear GTBP-1 stress granule formation. A combination of forward genetic screening together with RNAi-based candidate screening identified the P body components, the RNA helicase LAF-1, the SUMO protein SMO-1, and the mTOR pathway effector RSKS-1(S6K) as key regulators of GTBP-1 germline stress granule formation. Together, this work revealed that germline stress granules may be subjected to multiple layers of regulations and GTBP-1 may safeguards reproductive homeostasis under temperature stress by coordinating germline stress granule condensation. Author SummaryReproduction is particularly vulnerable to environmental stress, yet germ cells must remain functional to ensure fertility. How germ cells protect themselves under stressful conditions is still not well understood. In this study, we used the nematode Caenorhabditis elegans to examine how germ cells respond to heat stress, focusing on GTBP-1, a conserved protein involved in stress granule formation. We found that GTBP-1 plays contrasting roles depending on temperature: under normal conditions it restrains reproduction, whereas under heat stress it becomes essential for maintaining fertility. When animals are exposed to elevated temperature, GTBP-1 rapidly forms granules around the nuclei of germ cells, these granules assemble at sites occupied by processing bodies, another type of RNA-containing structure, revealing a close spatial relationship between two stress-responsive compartments. We also found that this process is regulated by conserved factors involved in RNA regulation and nutrient-responsive signaling. Together, our findings show how germ cells reorganize RNA-protein assemblies to preserve reproductive capacity under stressful conditions.

genetics↗

bioRxiv2026

Developmental Elimination of Electrical Synapses by UNC-51/UNC-76-Mediated Vesicular Transport

Neural circuit maturation requires precise elimination of transient synaptic connections, but how electrical synapses are eliminated remains unknown. Using C. elegans PLM mechanosensory neurons, we found that gap junctions form transiently during early development and are subsequently eliminated as the circuit matures. These transient electrical synapses drive high-frequency calcium oscillations, and optogenetic ablation of early gap junctions abolishes calcium dynamics and disrupts chemical synapse formation, demonstrating their critical developmental function. Through forward genetic screening, we identified the conserved kinase UNC-51/ULK as essential for elimination. UNC-51 phosphorylates the kinesin adaptor UNC-76/FEZ to initiate RAB-10-dependent retrograde trafficking, progressively depleting innexin proteins from the synaptic site. Disrupting this pathway prevents elimination, causing persistent electrical synapses and neuronal hyperactivity. These findings establish a conserved framework for developmental elimination of functionally essential transient electrical connectivity.

cell biology↗

bioRxiv2026

How wind and currents shape the drift velocity of macrophytes and macroplastic particles - from experiment to model

The post-detachment drifting phase of macrophytes, during which they can be alive, dead, or senescent, plays a crucial ecological and biogeochemical role by influencing long-range dispersal, transporting rafting species, affecting carbon sequestration, promoting blooms, and leading to beaching events. In order to predict the dispersal of macrophytes and macroplastic particles and where they will affect the ecosystem, it is important to be able to model how their drift velocities are influenced by hydrodynamic and aerodynamic factors. In this study, we investigated the drift velocity of macrophytes with diverse morphologies and macroplastic particles in a racetrack flume under different current conditions, in combination with and without wind in the same direction as the water current. Our data show that the drift velocity of macrophytes is highly dependent on their buoyancy and affected by morphological characteristics. Wind increased the velocity of the surface water, which in turn increased the drift velocity of both macrophytes and macroplastic particles. However, wind-induced turbulences reduced the overall effect, especially for macrophytes, which protruded minimally above the water surface in comparison to macroplastic particles. For positively buoyant specimens, an existing particle model was experimentally confirmed to predict macrophyte and macroplastic particle drift velocities reliably, irrespective of shape. For negatively buoyant species, we propose a novel equation to predict drift velocity, incorporating the diverse shapes of macrophytes, as well as their interaction with the bottom. These results represent the first step toward the development of trait-based models that represent macrophytes more realistically in dispersal simulations. O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=135 SRC="FIGDIR/small/709487v1_ufig1.gif" ALT="Figure 1"> View larger version (54K): org.highwire.dtl.DTLVardef@1ab9f6aorg.highwire.dtl.DTLVardef@6ef75dorg.highwire.dtl.DTLVardef@132334forg.highwire.dtl.DTLVardef@c6a3d8_HPS_FORMAT_FIGEXP M_FIG C_FIG

biophysics↗

bioRxiv2026

Mitochondria Organize Gap Junctions to Enable Neural Circuit Scaling

Neural circuits must maintain function during dramatic growth-driven expansion. While chemical synapses can be independently reorganized in pre- and postsynaptic neurons, gap junctions physically couple partner cells, requiring coordinated spatial redistribution--a process whose mechanisms remain unknown. Using the C. elegans mechanosensory circuit, we show that gap junctions undergo stereotyped developmental reorganization, transitioning from clustered to dispersed configurations as partner neurites elongate. Forward genetic screening identified RIC-7, a mitochondrial adaptor protein, as essential for this redistribution. Loss of RIC-7 prevents gap junction dispersal, trapping partner neurites at clustered sites and severely impairing electrical transmission. Mechanistically, RIC-7 directs mitochondrial positioning to gap junctions, where mitochondria recruit the microtubule-organizing component PTRN-1 to locally enhance microtubule dynamics. Disrupting mitochondrial transport (miro-1;mtx-2 mutants) or PTRN-1 function phenocopies ric-7 defects, while metabolic dysfunction does not, establishing a non-metabolic role for mitochondria in organizing synaptic architecture. Remarkably, RIC-7 expression in one neuron alone suffices to restore gap junction and partner neurite distribution across both coupled cells. Structural modeling suggests mechanical adhesion between docked hemichannels enables this coordinated reorganization. These findings establish mitochondria as mobile organizers of synaptic spatial architecture, demonstrate that proper junction distribution is essential for electrical transmission efficacy, and reveal how cell-autonomous mechanisms can achieve bilateral structural coordination during circuit growth.

neuroscience↗

bioRxiv2026

Lung-Selective Immune Reprogramming via In Situ Red Blood Cell Hitchhiking Nanoparticles

Premature clearance and limited organ targeting remain major barriers for nanoparticle (NP) drug delivery. Hitchhiking NPs on red blood cells (RBCs) can enhance circulation and organ-selective accumulation, but most approaches require ex vivo RBC extraction and reinfusion, limiting clinical translation. Here, we report an in situ RBC-hitchhiking strategy, named i-Bind, which employs polyphenol surface functionalization to enable spontaneous NP attachment to RBCs directly in the bloodstream. Driven by strong interactions of phenolic motifs with RBC membranes, i-Bind NPs exhibited markedly enhanced and more stable hitchhiking onto RBCs under flowing whole blood conditions. In both healthy and diseased mice, i-Bind NPs preferentially target the lungs, resulting in an over 20-fold increase in lung-to-liver deposition ratio compared to unmodified NPs. Additionally, i-Bind NPs show preferential targeting to distinct lung immune cell subsets in a pathology-dependent manner, including cDC2s in healthy lungs, neutrophils in acute lung injury, and cDC1s in lung metastases. In a melanoma lung metastasis model, delivery of the STING agonist diABZI via i-Bind NPs significantly inhibited lung metastasis progression by reprogramming the lung immune microenvironment. Collectively, i-Bind provides a simple and versatile platform for organ-targeted drug delivery and immune reprogramming. TeaserSurface functionalization of nanoparticles enables in situ red blood cell hitchhiking, unlocking new paths for organ-selective immune reprogramming

bioengineering↗

bioRxiv2026

MAP: A Knowledge-driven Framework for Predicting Single-cell Responses for Unprofiled Drugs

Predicting how cells respond to chemical perturbations is one of the goals for building virtual cells, yet experimentally profiled compounds cover only a small fraction of this space. Existing models struggle to generalize to unprofiled compounds, as they typically treat drugs as isolated identifiers without encoding their mechanistic relationships. We present MAP, a framework that integrates structured biological knowledge into cellular perturbation modeling and supports zero-shot prediction for small molecules with scarce or absent perturbation profiles. Specifically: (i) we construct MAP-KG, a large-scale knowledge graph tailored for cellular perturbation modeling that unifies 14 public resources, spanning 187k drugs, 23k genes, and 694k mechanistic relationships; (ii) we propose a knowledge-driven pre-training strategy that aligns molecular structures, protein sequence features, and textual mechanistic descriptions into a unified embedding space via contrastive learning, producing mechanism-aware and transferable gene and compound embeddings. The resulting knowledge-informed gene and drug representations are then coupled with a pretrained single-cell foundation model to condition perturbation response prediction; (iii) we evaluate MAP under two zero-shot generalization regimes: unseen cell type-drug combinations and the stricter setting of unprofiled drugs, where it improves top-50 DEG Pearson delta correlation by up to +13.3% and +12.2%, respectively, over the strongest baselines across three benchmarks. We further perform pathway-level functional analysis via GSEA for in-silico screening, where MAP predicts coherent, mechanism-consistent programs on unprofiled candidate drugs, and prioritizes 4 of 5 approved anti-cancer drugs in A-549 (non-small cell lung cancer).

bioinformatics↗

bioRxiv2026

PantheonOS: An Evolvable Multi-Agent Framework for Automatic Genomics Discovery

The convergence of large language model-powered autonomous agent systems and single-cell biology promises a paradigm shift in biomedical discovery. However, existing biological agent systems, building upon single-agent architectures, are narrowly specialized or overly general, limiting applications to routine analyses. We introduce PantheonOS (https://PantheonOS.stanford.edu), an evolvable, privacy-preserving multi-agent framework designed to reconcile generality with domain specificity. Critically, PantheonOS enables agentic code evolution, allowing evolving state-of-the-art batch correction and our reinforcement-learning augmented gene panel selection algorithms to achieve super-human performance. PantheonOS drives biological discoveries across systems: uncovering asymmetric paracrine Cer1-Nodal inhibition in proximal-distal axis formation of novel early mouse embryo 3D data; integrating human fetal heart multi-omics with whole-heart data to reveal molecular programs underpin heart diseases; and adaptively selecting virtual cell models to predict cardiac regulatory and perturbation effects. Together, PantheonOS points towards a future where scientific discoveries are increasingly driven by self-evolving AI systems across biology and beyond. Websitehttps://pantheonos.stanford.edu Ecosystemhttps://github.com/aristoteleo SummaryLarge language model-powered agent systems are driving a paradigm shift in scientific discovery by automating, scaling, and accelerating data analysis. This transformation is particularly profound in biology, where the rapid expansion of single-cell and spatial genomics has effectively reshaped the field into a data-intensive science. However, existing biological agent systems are typically constrained to single-agent designs, are narrowly specialized, or are overly general without sufficient domain expertise, limiting their applicability to routine or shallow analyses. Here, we introduce PantheonOS (https://pantheonos.stanford.edu), an evolvable, privacy-preserving, and general-purpose multi-agent framework designed to reconcile generality with deep domain specificity. PantheonOS provides an abstract, extensible architecture that enables customized agent composition and supports end-to-end single-cell and multi-omics analysis, spanning reinforcement-learning-augmented gene panel design, raw FASTQ processing, multimodal data integration, and three-dimensional spatial genomics reconstruction. Central to this framework, Pantheon-Evolve enables agentic code evolution, allowing the system to autonomously improve state-of-the-art batch-correction algorithms and new reinforcement-learning based gene panel design algorithms, achieving performance beyond manually designed baselines. We demonstrate the power of PantheonOS across multiple biological domains. In early mouse embryogenesis, PantheonOS automatically reconstructs three-dimensional spatial gene expression landscapes and resolves asymmetric Cer1 expression and paracrine Cer1-Nodal inhibition, revealing a robust proximal-distal axis at embryonic day six (E6.0). In human development, PantheonOS integrates fetal heart single-cell multi-omics with whole-heart 3D MERFISH+ data at post-conception week 12, uncovering spatially resolved molecular programs underlying heart disease ontogeny. Finally, an intelligent model-routing mechanism enables PantheonOS to adaptively select optimal virtual cell models across heterogeneous tasks, revealing minimal regulatory networks of cardiogenesis and predicting spatially resolved perturbation effects in the developing heart. Together, PantheonOS establishes a foundation for fully automated, evolvable, and domain-aware agentic analysis in genomics, and points toward a future in which scientific discovery is increasingly driven by self-improving AI systems across biology and beyond.

bioinformatics↗

bioRxiv2026

Leveraging Large Language Models to Extract Prognostic Pathology Features in Ewing Sarcoma

BackgroundCurrent risk stratification for Ewing sarcoma relies heavily on clinical factors such as metastatic status, failing to capture histologic heterogeneity as a potential prognostic indicator. Although pathology reports contain rich biological data, this information remains locked in unstructured narrative text, limiting large-scale retrospective analyses. We aimed to validate the utility of Large Language Models (LLMs) for scalable data abstraction and to identify prognostic histologic features from a large multi-institutional cohort. MethodsWe conducted a retrospective cohort study using data from six Childrens Oncology Group (COG) clinical trials. We utilized an LLM-based pipeline (OpenAI o3) to extract structured variables--including immunohistochemical (IHC) markers and CD99 staining patterns--from digitized, Optical Character Recognition (OCR)-processed pathology reports. Extraction accuracy was validated against a human-annotated ground truth (n=200) and cross-validated against senior experts (n=48). We assessed the association between extracted features and Overall Survival (OS) using Kaplan-Meier analysis and multivariable Cox proportional hazards regression, adjusting for metastatic status. FindingsWe analyzed 931 diagnostic pathology reports from 185 institutions spanning over 21-years. The LLM achieved a weighted average accuracy of 94% across 17 IHC markers; in a cross-validation subset, the LLM outperformed human annotators (weighted average accuracy over 15 IHC markers: LLM o3: 98.1%, a pediatric resident 91.4%, and a pediatric oncologist 95.9%). Survival analysis identified Neuron-Specific Enolase (NSE) and S100 as significant prognostic biomarkers. After adjusting for metastatic status, NSE positivity was associated with significantly inferior survival (HR 2.15, 95% CI 1.15-4.02, p=0.016); this risk was most pronounced in patients with non-metastatic disease (HR 5.64, p=0.0055). Conversely, S100 positivity was associated with improved survival (HR 0.58, 95% CI 0.34-1.00, p=0.046). InterpretationLLM-assisted extraction of pathology variables is highly accurate and scalable, capable of unlocking "dark data" from historical clinical trials. We identified NSE as a potent risk factor and S100 as a protective marker in Ewing sarcoma, particularly in localized disease. These findings suggest that AI-derived histologic data can refine risk stratification and, if validated, warrant inclusion in future prospective trials. Research in ContextO_ST_ABSEvidence before this studyC_ST_ABSWe searched PubMed and Google Scholar for articles published up to December 2025, using terms such as "Ewing sarcoma", "prognosis", "risk stratification", "large language models", "natural language processing", and "pathology report extraction". Current risk stratification for Ewing sarcoma relies predominantly on clinical variables, specifically the presence of metastatic disease and tumor size. While histologic heterogeneity is well-documented, it is rarely incorporated into risk models because extracting structured data from narrative pathology reports is labor-intensive. Small, single-institution studies have suggested potential prognostic roles for markers like Neuron-Specific Enolase (NSE) or S100, but results have been inconsistent and limited by small sample sizes. Furthermore, while Large Language Models (LLMs) have demonstrated potential for extracting clinical data, their application to "rescue" data from legacy, multi-institutional clinical trial documents (scanned PDF images) for rare disease biomarker discovery has not been validated at scale. Added value of this studyTo our knowledge, this is the largest study to utilize AI to extract histologic data from Ewing sarcoma pathology reports, aggregating 931 patients from six distinct Childrens Oncology Group (COG) clinical trials spanning over 21 years. We validated an LLM-based pipeline that converted noisy, optically character-recognized text from diverse institutions into structured data with 98.1% accuracy, outperforming human annotators. This scalable approach revealed that NSE positivity is a significant independent risk factor for mortality (HR=2.15), particularly in patients with non-metastatic disease where it confers a more than 5-fold increased risk of death. Conversely, we identified S100 positivity as a protective factor associated with improved survival (HR=0.58). Implications of all the available evidenceThis study demonstrates that LLMs can reliably unlock "dark data" from historical clinical trials, rendering vast archives of unstructured medical documents accessible for retrospective analysis. The identification of NSE and S100 as robust prognostic biomarkers suggests that these widely available immunohistochemical stains provide valuable information beyond standard diagnostic information. These findings support the integration of automated data extraction tools in research workflows and suggest that NSE and S100 status should be considered in the design of future risk-stratified clinical trials for Ewing sarcoma.

bioinformatics↗

bioRxiv2026

The genetic repertoire of the deep sea: from sequence to structure and function

The deep sea as the largest and maybe most hostile environment on Earth is still underexplored especially regarding its genetic repertoire. Yet, previous work has revealed significant habitat-specific deep-sea biodiversity. Here, we present an integrated deep-sea genetic dataset comprising 502 million nonredundant genes from 2,138 samples and 2.4 million predicted structures, and used it to link specific protein structures with genetic variants associated with life in the deep sea and to assess their biotechnology potential. Combining global sequence analysis with biophysical and biochemical measurements revealed unprecedented sequence diversity, yet substantial structural conservation of proteins. Especially proteins involved in replication, recombination, and repair were identified to be under rapid evolution and with specialized properties. Among these, a structurally divergent helicase exhibited advantages in controlling nanopore sequencing speed. Thus, our work positions the deep sea as a unique evolutionary engine that generates and hosts genetic diversity and bridges genetic knowledge with biotechnology.

bioinformatics↗

bioRxiv2026

Single-cell Multiome Analysis of Chromatin State and Transcriptome in the Human Basal Ganglia

The basal ganglia play essential roles in motor control, emotion, learning and reward processing. Their dysfunction contributes to many neurological and psychiatric disorders. However, the gene regulatory programs defining basal ganglia cell-type identity and function remain poorly understood, limiting interpretation of disease-associated non-coding variants. Here, we present the first single-cell multiome atlas of histone modifications and transcriptomes across eight basal ganglia regions from neurotypical adult human donors. Joint profiling reveals cell-type-specific deployment of active and repressive cis-regulatory elements and gene regulatory networks, and suggests a combinatorial homeobox transcription factor code underlying cell identity. Integration with matched spatial transcriptomic MERFISH data uncovers regional heterogeneity of epigenomic landscapes. Comparative analysis between human and mouse medium spiny neurons uncovers conservation of core gene regulatory features. This atlas interprets non-coding risk variants of neuropsychiatric disorders and supports the development of a deep learning model to predict gene regulation and functional effects of disease-associated variants. HIGHLIGHTSO_LIJoint single-cell profiling of transcriptomes and three histone modifications across eight human basal ganglia regions characterizes active and repressive chromatin states at cell-type resolution. C_LIO_LICell-type-specific gene regulatory programs decode combinatorial homeobox TF grammar governing the identity and diversification of basal ganglia neurons. C_LIO_LIIntergrative analyses link noncoding neuropsychiatric risk variants to specific cell types, regulatory elements, and candidate target genes. C_LIO_LIA sequence-to-function deep-learning model predicts gene regulation from DNA sequence and prioritizes functional disease-associated variants. C_LI

genomics↗