Biology subjects

Yang, Q.

Publications and source records attributed to Yang, Q..

At least 91 records · Page 5Linked to original sources

bioRxiv2025

Aging-induced hepatocyte CD44 drives IL6/STAT3 signaling and associates with impaired neighboring T cell function.

Liver cancer incidences increase dramatically beyond 55 years of age, suggesting that age-associated changes contribute critically to tumor initiation. However, the mechanisms linking liver aging and cancer initiation are not well defined. This study investigates the role of CD44, a marker of liver tumor-initiating cells (TIC), in age-associated liver pathophysiology. Aged livers showed accumulation of CD44-expressing hepatocytes exhibiting enrichment of immune modulatory genes and activation of the immunosuppressive IL6/JAK/STAT3 pathway. Indeed, in adoptive transfer assays, antigen-exposed CD8+ T cells mounted a lower IFN-{gamma} response in aged livers than in young livers, indicating an immunosuppressive aged milieu. Concordantly, spatial analyses showed that the proximal neighbourhoods of Cd44-expressing hepatocytes are enriched in T cells exhibiting reduced cytokine and chemokine gene expression. Finally, hepatocyte-specific knock out of Cd44 mitigated the IL6/JAK/STAT3 gene signature in aged livers. Overall, these findings suggest that CD44 expression in aged hepatocytes promotes activation of the immunosuppressive IL6/JAK/STAT3 pathway and this is associated with impaired T cell effector function.

cell biology↗

bioRxiv2025

HINS: HLA-mediated immunogenic neoantigen score for robust prediction of immune checkpoint blockade response

Immune checkpoint blockade (ICB) therapy response prediction provides remarkable genomic and transcriptomic gains and has successfully understood the molecular mechanisms underlying various cancers. However, only a subset of patients with advanced tumors currently benefits from ICB therapies, which at times incur considerable side effects and costs. Developing predictive tools for ICB response has remained a serious challenge because of the complexity of the immune response and the shortage of large cohorts of ICB-treated patients that include both omics and response data. Here, based on a pooled genomic dataset of 919 patients across multiple solid tumors treated with anti-PD-(L)1 or anti-CTLA-4, we constructed human leukocyte antigen class I (HLA-I)-mediated immunogenic neoantigen score (HINS). HINS is a predictor of ICB response using neoantigen quality and HLA evolutionary divergence to integrate the factors associated with immune activation and evasion. It yielded an overall accuracy of AUC=0.853, outperforming existing predictors and capturing more than 75% true responders. Experimental analysis indicated that patients with higher HINS were more likely to undergo survival benefits following ICB therapy. Our results highlighted the transcriptional and genomic correlations between HINS-identified molecule features and ICB response, such as immunoresponsive gene enrich pathways, favorable and unfavorable genomic subgroups, and hotspot somatic events in driver genes. This study presented an interpretable, accurate deep-learning method using meaningful predictive features to predict the response to ICB, providing biological insights into the complexity of the determinants underlying immunotherapy.

bioinformatics↗

bioRxiv2025

Integration of single-cell transcriptome and genetic profiles reveals critical cell types and genes in inflammatory bowel disease

The progression of inflammatory bowel disease (IBD), grouped into Crohns disease (CD) and ulcerative colitis (UC), involves multiple cell types and genes, but the identification of critical cell types and genes that contribute to intestinal inflammation remains challenging. By integrating human intestinal single-cell transcriptomic data from approximately 200 donors, we revealed the stromal cell-derived network in CD, enrichments of immune cells such as Th1 in CD, atypical B cells and Tfh in UC, AXL+SIGLEC6+ DCs and IgG+ Plasma B cells in both, and the association between SPP1+ Macrophages, FAP+ Fibroblasts and anti-TNF treatment resistance. We also investigated the gene expression characterization of enteric neurons. Additionally, by linking cell-type-specific transcriptional alterations with genetic profiles of disease risk and intestinal gene expression, we proposed B cells and cycling epithelial cells as critical cell types, and identified epithelial genes, such as GPX4, as associated with CD pathology. In summary, our study comprehensively characterized the single-cell landscape of human intestines in IBD and provided insights into cell types and genes highly involved in genetically determined CD pathology, thereby informing CD prediction, diagnosis, and the development of new therapeutic targets.

immunology↗

bioRxiv2025

Deep-learning model guided discovery and characterization of bacterial unspecific peroxygenases

Unspecific peroxygenases (UPOs) are capable of catalyzing the selective oxidation of organic substrates under mild conditions, using hydrogen peroxide (H2O2) as the sole oxidant. This makes them one of the most promising biocatalysts for chemical synthesis. However, the major limitation restricting the application of UPOs to date is their difficulty in heterologous expression. Although more than 4,000 putative UPO enzymes have been recorded in databases, only about 50 of them can currently be heterologously expressed. All UPOs discovered so far originate from eukaryotes (mainly basidiomycetes and some ascomycetes), and they rely on the complex expression systems and post-translational modifications of their native hosts. This further exacerbates the challenges associated with heterologous expression of eukaryotic UPOs. In this work, we developed a deep-learning-based enzyme mining strategy and, for the first time, discovered novel UPO enzymes from bacteria, achieving successful heterologous expression in Escherichia coli. Bacterial UPOs differ greatly from fungal UPOs in sequence similarity, displaying completely distinct evolutionary trajectories. The discovery of bacterial UPOs advances our understanding of the catalytic mechanisms and expression characteristics within the UPO family, breaking the long-held assumption that UPOs can only originate from eukaryotes. Their excellent heterologous expression performance and broad catalytic versatility will further expand the application potential of UPOs.

bioinformatics↗

bioRxiv2025

Ferrapoptosis: Discovery of a Hybrid Programmed Cell Death

Traditional taxonomies classify ferroptosis and apoptosis as distinct forms of regulated cell death. Here, we challenge this view by defining and validating a previously unrecognized neuronal death entity in traumatic brain injury (TBI), which we term "Ferrapoptosis". This hybrid mode of death is characterized by the coexistence of ferroptotic and apoptotic molecular and ultrastructural features within the same neuron-- ultrastructural alterations that cannot be fully characterized by any known cell death modalities. By integrating single-cell transcriptomics with multi-layered functional assays, we show that Ferrapoptosis dominates in severe injury and the acute phase, and is critically driven by mitochondrial oxidative stress. Over time, its predominance is gradually replaced by death modes in which either ferroptosis or apoptosis alone becomes the major pathway. Genome-wide CRISPR screening further identifies Smg7 as a key regulator that synchronously activates both death programs to drive the hybrid phenotype, whereas genetic or viral inhibition of Smg7 reduces Ferrapoptosis and promotes neurological recovery after TBI in mice. Our work systematically delineates a previously unrecognized form of cell death and its pathogenetic mechanisms, providing experimental evidence to refine cell death classification, and suggesting a conceptual strategy for treating complex diseases such as TBI by targeting shared regulatory nodes. HIGHLIGHTSO_LIDefined and validated a previously unrecognized cell death modality, termed "Ferrapoptosis". C_LIO_LIElucidated the ordered pattern of Ferrapoptosis across key stages of TBI pathogenesis. C_LIO_LIRevealed mitochondrial oxidative stress as the pivotal hub integrating ferroptosis and apoptosis. C_LIO_LIIdentify Smg7 as a core driver of Ferrapoptosis and a promising neuroprotective target. C_LI

cell biology↗

bioRxiv2025

Inherited retinal degenerations: PARP regulates calpain activation via TRPM2 channels in rd1 mouse photoreceptors

Inherited retinal degeneration (IRD) refers to untreatable blinding diseases characterized by progressive photoreceptor loss. Photoreceptor degeneration is often associated with an excessive activation of poly (ADP-ribose) polymerase (PARP) and Ca2+-dependent calpain-type proteases. To explore the interplay between PARP and calpain activity, we employed organotypic retinal explant cultures derived from wild-type mice and from the rd1 mouse model for IRD. Retinae were treated with the PARP inhibitors INO1001 or Olaparib, the poly (ADP-ribose) glycohydrolase (PARG) inhibitor JA2131, or the transient receptor potential channel M2 (TRPM2) blocker 8-Br-ADPR. Readouts included the TUNEL assay to detect cell death, in situ activity assays for histone-deacetylases (HDAC), PARP, and calpain, as well as immunostaining for activated calpain-2, and poly (ADP-ribose) (PAR). PARP, PARG, and TRPM2 inhibition reduced calpain activity and calpain-2 activation. PARP activity was decreased by PARP and TRPM2 inhibitors but not by PARG inhibition. Remarkably, the PARP inhibitor INO1001 increased HDAC activity unlike any of the other compounds. When combined with the PARG inhibitor JA2131, INO1001 reduced photoreceptor cell death in a synergistic fashion, although such synergy was not observed for calpain or PARP activity. Moreover, synergistic photoreceptor preservation was not observed when JA2131 was combined with the PARP inhibitor Olaparib. Overall, these results indicate that in rd1 photoreceptors, PARP controls calpain activity via PARG and TRPM2-induced Ca2+ influx. We also characterize INO1001 as potentially more beneficial for IRD treatment than Olaparib. Our study details the complexity of PARP-signalling in photoreceptors and identifies PARG and TRPM2 as new targets for IRD therapy development.

neuroscience↗

bioRxiv2025

Dual-Channel Event Microscopy for Ultrafast Biological Imaging

Many fundamental biological processes--spanning immune-tumor interactions, neuronal signaling, and microvascular flow--exhibit fast, multiscale dynamics among diverse cell types within three-dimensional tissue environments. Capturing such activity requires imaging systems that simultaneously achieve high temporal resolution, multicolor capability, and volumetric coverage over large fields of view (FOVs). However, existing modalities remain limited by trade-offs among imaging speed, spectral capacity, depth of field (DOF), and spatial resolution. Here, we present Dual-Channel Event Microscopy (DEM), which integrates digital micromirror device (DMD)-based pulsed illumination, extended-DOF optics, and event-based sensing for ultrafast, dual-channel volumetric imaging across a 2.3 x 1.3 mm2 FOV with an effective 200 {micro}m DOF. Using dual-color fluorescent phantoms and microsphere flow assays, DEM achieves accurate spectral separation and reconstruction of rapid motion at kilohertz frame rates. In vivo, DEM enables simultaneous visualization of neutrophils and premalignant tumors in freely swimming zebrafish. In immobilized specimens, it provides robust, sensor-level optical sectioning near the heart, suppressing diffuse background to reveal fine vascular networks and active blood circulation into and out of the cardiac chambers. DEM further enables quantitative mapping of blood-flow dynamics in the zebrafish tail, resolving arterial-venous differences and capturing heartbeat-driven oscillations that reflect cardiac pumping with high temporal fidelity. By uniting ultrafast acquisition, dual-channel capability, volumetric coverage, and intrinsic optical sectioning within a single event-driven architecture, DEM offers a powerful platform for visualizing rapid multicellular interactions and physiological dynamics in living systems.

bioengineering↗

bioRxiv2025

DNA Framework Nanoreactor for Programmable Membrane Fusion

Membrane fusion is a fundamental yet transient process that has long resisted direct structural and kinetic dissection. Here we introduce a DNA framework vesicle (DFV) nanoreactor that transforms this elusive biological phenomenon into a programmable, visualizable, and quantifiable process-a nanoreaction confined in space yet extended in time. By embedding lipid membranes and SNARE proteins within precisely defined DNA apertures, DFVs convert stochastic vesicle collisions into geometry-controlled fusion events with tunable kinetics. Cryo-electron microscopy resolves a complete sequence of six intermediates, revealing how nanoscale confinement reshapes the energetic landscape of bilayer merger. Quantitative fluorescence and nano-flow cytometry further establish a direct link between spatial design and fusion probability. Extending this concept to living cells, DFVs enable controllable membrane fusion and augmentation on VAMP2-expressing membranes, achieving direct cytosolic delivery of functional siRNA via fusion-driven transfer rather than endocytosis. This framework bridges structural precision and functional mimicry, offering a unified platform to reconstruct, quantify, and harness membrane fusion as a programmable process for synthetic biology and nanomedicine.

bioengineering↗

bioRxiv2025

PGC-1α and PPARs cooperatively mediate photoreceptor neuroprotection in rd1 mouse inherited retinal degeneration

Retinitis pigmentosa (RP) is a group of inherited diseases characterized by a primary rod photoreceptor dysfunction and progressive rod and cone cell death. Due to their very high energy demand, the degeneration of photoreceptors may be linked to insufficient energy supply or metabolic imbalance. Critical transcription factors that regulate metabolism such as peroxisome proliferator-activated receptors (PPARs) and their co-activator PGC-1 have been found to play important roles in neurodegenerative diseases, but their potential roles in RP have yet not been disclosed. In this study, we used organotypic retinal explant cultures derived from the rd1 mouse model for RP to investigate the effects of PPAR, PPAR{gamma}, PPAR{beta}/{delta} agonists, as well as PGC-1 activation and inhibition. Photoreceptor death in the outer nuclear layer (ONL) of the retina was quantified using the TUNEL assay, while in situ activity assays were used to monitor effects of PPARs and PGC-1 on poly (ADP-ribose) polymerase (PARP) and calpain activity. In addition, we performed immunostainings to evaluate poly (ADP-ribose) (PAR) generation and activation of calpain-1 and calpain-2. We found that PPAR{beta}/{delta} agonists had limited effects, while activation of PPAR, PPAR{gamma}, and PGC-1 significantly reduced photoreceptor death and PARP activity in rd1 retina. Conversely, inhibition of PGC-1 had a strong detrimental effect on photoreceptor viability. Activation of the histone deacetylase sirtuin-1, an upstream agonist of PGC-1, had no effect unless it was combined with simultaneous inhibition of PARP. Furthermore, PPAR{gamma} and PGC-1 effectively suppressed overall calpain activity and overactivation of calpain-2, alleviating photoreceptor degeneration caused by Ca2+ imbalance. In summary, our data supports the concept of a PARP-sirtuin-1-PGC-1-PPAR-PARP feedback control that connects defective energy metabolism to photoreceptor degeneration. Specifically, our findings suggest that PPAR, PPAR{gamma}, and PGC-1 cooperate to preserve photoreceptor viability, highlighting PPAR-signaling as a promising target for future therapeutic interventions.

neuroscience↗

bioRxiv2025

Targeting the host factor HGS-viral membrane protein interaction in coronavirus infection

While current antivirals primarily target viral proteins, host-directed strategies remain underexplored. Here, we performed a genome-wide CRISPRi screening to identify the host protein, Hepatocyte Growth Factor-Regulated Tyrosine Kinase Substrate (HGS), as essential for the pan-coronaviruses infection both in vitro and in vivo. Mechanistically, HGS directly interacts with the viral membrane (M) protein, facilitating its trafficking to the ER-Golgi intermediate compartment (ERGIC) for virion assembly. Conversely, HGS deficiency caused M retention in the ER, blocking assembly. Leveraging this interaction, we designed M-derived peptides and screened over 5,000 FDA-approved drugs, identifying riboflavin tetrabutyrate (RTB). Both the peptides and RTB bind HGS and disrupt its interaction with the M protein, leading to M retention in the ER and subsequent blockade of virion assembly. These agents demonstrated broad anti-pan-coronavirus activity in vitro and in vivo. Collectively, our findings establish HGS as a druggable host target and identify RTB as a promising broad-spectrum antiviral candidate.

microbiology↗

bioRxiv2025

Loss of tight junction protein claudin 18 uncovers alveolar epithelial stem cell plasticity and emergence of non-fibrogenic transitional progenitors

Persistence of senescent alveolar transitional progenitors following lung injury is implicated in the pathogenesis of fibrosis. We identified transitional cells in uninjured Cldn18 knockout (KO) mouse lungs distinct from previously reported damage-associated transitional progenitors (DATPs) with a less fibrogenic transcriptomic profile. Cldn18 KO mice are protected from bleomycin-induced fibrosis, with early restoration of cellular homeostasis. Lineage tracing implicates accelerated differentiation as a mechanism for protection from fibrosis, leading us to name these cells regeneration-associated transitional progenitors (RATPs). Multiome confirms that RATPs and DATPs are epigenetically distinct, with RATPs comprised of RATP2s and RATP1s based on epigenomic proximity to AT2s and AT1s, respectively, and suggests dynamic regulatory remodeling during AT2-to-AT1 differentiation, with NKX2.1 and AP-1 active in early transitions and TEAD factors in later stages. These results reveal an unexpected role for Cldn18 in regulation of AEC plasticity, while identification of RATPs challenges the notion that persistence of transitional alveolar cells is invariably pathologic.

cell biology↗

bioRxiv2025

Host constraints on viral recombination and emergence

Recombination (including reassortment) is a salient force in viral evolution and has been implicated in the emergence of several zoonotic pathogens in humans. Viral recombination occurs during simultaneous infection of an individual host with multiple genotypes (co-infection). Thus, processes which affect the incidence of a disease in the host population affect how often viral genotypes recombine. We investigate whether and how host traits affect the realized rate of viral recombination using a mathematical model that makes feedbacks between viral evolution and host ecology (in particular, lifespan) explicit. Our main result is that viruses of host species that are short-lived tend to recombine more frequently than those of relatively long-lived hosts. This is because of differences in population density and, thus, the prevalence of (co-)infection at equilibrium. Using highly pathogenic avian influenza sequence data, we test the prediction that recombination is elevated in short-lived hosts. In agreement with this prediction, the magnitude of statistical associations between mutations on different segments of the flu genome increases with host body size, a proxy for lifespan. Similarly, estimates of the reassortment rate from phylogenetic network analyses decrease with body size. We discuss the implications of these findings for disease emergence. Subject categoryEvolution Subject areasevolution, health and disease and epidemiology, ecology

evolutionary biology↗

bioRxiv2025

Unveiling the Chain Mediation Model of Sleep Quality and Fatigue in Linking Subjective Exercise Experience to Depression in Adolescent

While the positive correlation between physical exercise and depression has been widely replicated in cross-sectional research, the prognostic value of subjective exercise experience for depression continues to be disputed, and the mediating processes are not well understood. This study aims to elucidate the predictive role of subjective exercise experience in depression and examine the chain mediating effects of sleep quality and fatigue. We administered validated self-report questionnaires to assess these constructs in a sample of 538 Chinese high school students. The results demonstrated that adolescents depressive symptoms are influenced by subjective exercise experience, and this effect mediated through two independent pathways: sleep quality and fatigue. Moreover, subjective exercise experience can predict adolescent depression via either the sleep quality pathway or the fatigue pathway. The identification of mediating dual pathways provides empirical support for developing targeted exercise interventions that specifically address sleep disturbances and fatigue management in depressed adolescents.

physiology↗

bioRxiv2025

Vitamin D3 ameliorates R-loop-induced replication stress and chromosomal instability in MED12-mutant uterine fibroids

Uterine fibroids (UFs) are the most important benign neoplastic threat to womens health worldwide, with no long-term noninvasive treatment options currently available. Among known UF driver alterations, somatic mutations in Mediator subunit MED12 are by the far the most prevalent, accounting for up to 80% of these clinically significant lesions. Although it is presently unclear how MED12 mutations trigger neoplastic transformation, MED12-mutant UFs are nonetheless characterized by significant chromosomal loss and rearrangement, suggesting genomic instability as a driving force in tumor development. However, the basis by which MED12 mutations drive genomic instability is not known. Herein, we show that R-loop-driven replication stress in MED12-mutant UFs leads to DNA under-replication and mitotic segregation errors that drive chromosomal instability. Notably, we find that vitamin D3 (VD3), a modifiable risk factor in UF development, suppresses pathogenic R-loop accrual and ameliorates replication stress-driven chromosomal instability, contributing to growth inhibition of patient-derived MED12-mutant UF xenografts in vivo. Altogether these findings uncover a molecular basis by which the predominant UF driver converges with a known risk factor at the interface of genomic instability, with significant translational implications for personalized UF prevention and treatment.

molecular biology↗

bioRxiv2025

The Genetic Immune Basis of Gout: Identification, Functional Characterization, and Therapeutic Implications

Gout, the most common inflammatory arthritis, is still managed mainly as a metabolic disorder, with treatment centered around urate lowering rather than blocking the immune response that drives gout flares. This emphasis reflects a historical view of gout as a disease of excess and diet, reinforced by genome-wide association studies (GWAS) that consistently detect urate transport and metabolism loci, while immune-related loci have failed to reach genome-wide significance. Growing clinical and experimental evidence, however, points to a central role for immune pathways in gout pathogenesis. We set out isolate the immune genetic component of gout. Using a conjunctional false discovery rate (conjFDR) framework, we integrated gout GWAS data with those from eight immune-mediated disorders rheumatoid arthritis, Crohns disease, inflammatory bowel disease, psoriasis, multiple sclerosis, asthma, chronic obstructive pulmonary disease, and systemic lupus erythematosus. We observed robust pleiotropic enrichment across all comparisons and mapped the resulting loci through FUMA to 85 unique credible genes broadly distributed across the genome, with no classical urate transporters present. Additionally, we identified 16 novel genes for gout, many of which are of obvious immune nature. We performed GO and KEGG enrichment analyses, which at a stringent q-value threshold identified adaptive and innate immune pathways, including T and B cell receptor signaling, antigen processing and presentation, NF-{kappa}B signaling, and Jak-STAT signaling. At a nominal p-value threshold, we uncovered additional cytokine-driven processes such as IL6 and IL7 signaling. We confirmed disease relevance with DisGeNET and established causality for 14 genes through Mendelian randomization, including IL1RN, MAP3K11, and SH2B3 genes with existing pharmacological inhibitors. We are the first to genetically isolate the genetic immune component of gout. Our findings show that these immune pathways can be specifically targeted, and immune medication should be incorporated into therapeutic strategies to complement urate-lowering approaches.

immunology↗

bioRxiv2025

Resolving the estrogen paradox in hereditary retinal degeneration: Esr1 activation suppresses Tnf-α signaling as a photoreceptor self-protection mechanism

Retinitis pigmentosa (RP) is an inherited retinal degenerative disorder characterized by progressive photoreceptor loss and irreversible blindness. Increasing evidence implicates neuroinflammation as a contributor to photoreceptor degeneration extending beyond the initial genetic insult. Although estrogen has been reported to exert anti-inflammatory effects in the central nervous system, its role in RP remains controversial, with some studies suggesting a paradoxical exacerbation of retinal pathology. To address this discrepancy, we identify estrogen receptor alpha (Esr1) as a central immunoregulatory hub in RP. Transcriptomic analyses of rd1 and rd10 revealed upregulation of estrogen-responsive and inflammatory pathways, with Esr1 expression markedly elevated during degeneration. TUNEL assays demonstrated that systemic estradiol (E2) exerted divergent effects, protective in rd1 yet deleterious in rd10, whereas selective pharmacological activation of Esr1 with propyl pyrazole triol (PPT) consistently reduced photoreceptor death, preserved dark-adapted ERG responses, and downregulated inflammatory mediators including Tnf-, Cx3cl1/Cx3cr1, Cd68, and Iba1. Mechanistically, Esr1 activation repressed microglial Tnf transcription and disrupted a self-sustaining Cx3cl1/Cx3cr1-Tnf- signaling loop driving microglial recruitment, activation and neurotoxicity in the outer nuclear layer (ONL). Targeted interventions confirmed tumor necrosis factor receptor 1 (Tnfr1) as the principal mediator of Tnf-induced photoreceptor death: selective inhibition with R7050 conferred superior protection compared with broad-spectrum Tnf- inhibitors (etanercept, infliximab). Cx3cr1 blockade likewise suppressed microglial activation and improved visual outcomes. Collectively, our findings establish Esr1 activation as not merely an external intervention but the amplification of an intrinsic self-protective program, positioning Esr1, Tnfr1, and Cx3cr1 as actionable therapeutic targets to suppress neuroinflammation and preserve vision in RP and related retinal disorders.

neuroscience↗

bioRxiv2025

BioLab: End-to-End Autonomous Life Sciences Research with Multi-Agents System Integrating Biological Foundation Models

Scientific discovery in the life sciences remains hindered by fragmented workflows, narrow-scope computational models, and inefficient links between in silico prediction and wet-lab validation. We present BioLab, a multi-agent system that integrates domain-specialized foundation models to automate end-to-end biological research. BioLab comprises eight collaborating agents, including a Planner, Reasoner, and Critic, orchestrated through a Memory Agent that enables iterative refinement via retrieval-augmented generation and a suite of 219 computational xBio-Tools spanning five biological scales (DNA, RNA, protein, cell, and chemical). These tools are built on the xTrimo Universe, a collection of 104 models derived from six foundation models (xTrimoChem, Protein, RNA, DNA, Cell, and Text), the majority of which achieve state-of-the-art (91-100% SOTA ratios) on domain benchmarks. Across standard reasoning tasks (PubMedQA, MMLU-Pro/Biology, GPQA-diamond), BioLab consistently outperformed leading large language models, including GPT-4o, Gemini-2.5, and DeepSeek-R1. Beyond benchmarks, BioLab autonomously executed a fully computational pipeline for de novo macrophage-targeting antibody design, progressing from target mining to multi-objective antibody optimization, where molecular dynamics simulations revealed structural mechanisms underlying enhanced affinity of optimized variants. Closing the computational-experimental loop, BioLab designed optimized antibodies (Pem-MOO-1, Pem-MOO-2) that achieved IC50 values of 0.01-0.016 nM, markedly surpassing the parental Pembrolizumab (0.027 nM) for PD-1. Functional assays confirmed enhanced pathway blockade and improved multi-parameter performance profiles. Together, these results establish BioLab as a generalizable framework for AI-native scientific discovery, demonstrating how multi-agent systems coupled with foundation models can autonomously generate, execute, and experimentally validate novel biological hypotheses.

bioinformatics↗

bioRxiv2025

Large-language-model-based antibiotic resistance gene prediction and resistomes mining in cyanobacterial blooms

The rapid spread of antibiotic resistance genes (ARGs) in aquatic ecosystems poses a serious public health threat. Conventional ARG detection methods, including sequence-alignment and machine learning, are limited by high false-negative or false-positive rates, especially due to sequence diversity and class imbalance in metagenomic data. Moreover, few tools offer comprehensive solutions for ARG identification, classification, and functional annotation. To address these limitations, we developed ESMARG, a novel protein language model framework based on ESM1v, for accurate ARG identification, classification, and annotation. Trained with both ARG and abundant non-ARG sequences from real-world metagenomes to enhance robustness and reduce false positives, ESMARG significantly outperformed traditional alignment-based (BLAST, DIAMOND) and recent deep learning models (ARGNet, ARG-SHINE), achieving 0.998 precision, 0.939 recall, and 0.968 F1-score for ARG identification. Functional and mechanism classification modules also demonstrated high accuracy (0.986 and 0.987) and computational efficiency. We applied ESMARG to analyze 26 cyanobacterial aggregate (CA) metagenomes collected from Lake Taihu across a full annual cyanobacterial harmful algal bloom (CyanoHAB) cycle. A total of 110 ARGs, spanning 24 drug classes and multiple resistance mechanisms, were detected, with seasonal shifts in ARG abundance and composition. Strong correlations were identified between CA resistomes and microbial community structure, especially with cyanobacteria. Further, ARG abundance was shaped by both biotic factors--such as cyanobacterial dominance and bacterial functional profiles--and abiotic variables, including biochemical oxygen demand and water temperature. Our findings demonstrate the power of ESMARG for high-resolution ARG profiling and reveal complex ecological interactions between resistomes, microbiomes, and environmental factors in CyanoHAB. HighlightsO_LIDeveloped ESMARG model outperforms BLAST/other deep learning models in ARG detection C_LIO_LIARG showed seasonal dynamics in cyanobacterial metagenomes across algal bloom cycle C_LIO_LIARG was strongly correlate with microbiome structure and cyanobacteria dominance C_LIO_LIARG abundance was shaped by cyanobacteria, bacterial functions, BOD and temperature C_LI

bioinformatics↗