Biology subjects

Chen, L.

Publications and source records attributed to Chen, L..

At least 199 records · Page 11Linked to original sources

bioRxiv2025

Undersampling techniques for large datasets

DNA-Encoded Libraries allow for an efficient approach to synthesize and screen billions of small molecules against a target of interest. With more real-world binding data, this can improve training of machine learning models. However, one key challenge in DELs is the severe imbalances between the classes, in other words, there are much more inactive than active compounds against any given target. This can heavily skew the training process. In this study, we explore different undersampling strategies for the majority class. These different techniques are benchmarked against random selection and prototyped on two different DEL datasets with three different machine learning models. Overall, the max_sim strategy shows the best scores, and the general pipeline is implemented in the DELight package.

bioinformatics↗

bioRxiv2025

Reference-Informed Spatial Domain Detection Using Weak Supervision for Spatial Transcriptomics

One of the key objectives in spatial transcriptomics (ST) studies is to map the complex organization and functions of tissues. We introduce GraphScrDom, a reference-informed and weakly supervised contrastive learning model that uniquely integrates expert-provided manual annotations (i.e., scribbles) on spatial grids or histology images with cell type-specific gene expression profiles derived from reference single-cell RNA-seq data to perform tissue segmentation. With only limited scribble annotations, GraphScrDom consistently outperforms existing methods across various ST platforms and at both bulk and single-cell resolutions, as evaluated by six widely used metrics, demonstrating strong generalizability and robustness. Additionally, we have developed an integrative software toolkit that includes an interactive annotation interface and a model training module for spatial domain detection, providing a unified and user-friendly framework to facilitate spatial domain analysis.

bioinformatics↗

bioRxiv2025

Higher-order Residue Interactions Encode Diversified Physical Properties of Biomolecular Condensates

Biomolecular condensates often formed by proteins containing extended intrinsic disordered regions (IDPs) can coordinate versatile cellular processes. The biological functions of condensates are attributed to a variety of physical properties, including viscoelasticity, interfacial tension and saturation concentration. Comprehensive understanding of biological functions of protein condensates and the corresponding sequence encoding requires unified interpretation of the diverse physical properties by a single model. Here we report the diversified physical properties of condensate and the interactions between constituent IDPs can be interpreted on the basis of higher-order residue interactions. These findings are based on s-scale atomistic molecular dynamics simulations of three homogeneous IDP condensates (FUS PLD, ARF6 PLD, LAF-1 RGG) combined with the reported simulation trajectories of two heterogeneous condensates. In all interacting IDPs, there are residue interactions between finite motifs (3[~]5 residues) that organize in a highly cooperative way: residues form multivalent interactions and their remaking is localized. The sequence features of higher-order interactions (HOIs) are identified: tyrosine, arginine, glutamine and aspartic acid are essential, and their consecutive sequence position is also important. The interaction between corresponding motifs is highly sustained and dominates the cohesive energy of IDP. IDP condensate can be modeled as a HOI-crosslinked network. This model can successfully reproduce all reported experimental results of condensate viscoelasticity, interfacial tension and saturation concentration, and expand our insights into the molecular grammar of IDP interactions. Significance StatementThe biological functions of biomolecular condensates involve various physical properties of condensate. A unified model that describes all these physical properties is essential for the understanding of condensate functions. Our research reveals the prevalence of higher-order residue interactions in all IDP condensates we considered, which leads to extremely sustained interaction between corresponding residue motifs. Accordingly, we propose a physically crosslinked network based on higher-order residue interactions as a simplified model for condensate. This model can quantitively reproduce all reported experimental results of condensate physical properties, including viscoelasticity, interfacial tension and saturation concentration.

biophysics↗

bioRxiv2025

Chronic Cadmium Exposures and Hyperglycemia Additively Drive Mitochondrial Dysfunction in Hepatic Cells: Key Implications for MASLD Etiopathogenesis

Effects of chronic heavy metal stress on hepatocellular pathophysiology remains ill-understood. Human livers are a long-term accumulative site for many toxic heavy metals (e.g., cadmium and arsenic) whose effects are unknown. In the current study, we studied effects of chronic, low-dose exposures of cadmium (CLEC) modulated by normoglycemic (5.6 mM) and hyperglycemic (15 mM) exposures, focusing on hepatocellular mitochondrial function. HepG2 and HUH7 cell lines were exposed to CLEC and glucose for 24 weeks, mimicking a chronic heavy metal exposure paradigm seen in normal and type II diabetic individuals. We observe that CLEC exposures significantly affect the long-term health of mitochondria, including decreased mitochondrial mass, increased superoxide production, and loss of mitochondrial membrane potential (MMP) in a CLEC and glucose-dependent manner. Furthermore, the Seahorse MitoStress assay revealed CLEC induced significant chronic oxidative stress. In particular, CLEC cells showed altered levels of basal and non-mitochondrial respiration, causing dysregulation in mitochondrial oxygen consumption rates (OCRs). Lastly, we identified significant impacts of CLEC and glucose exposures on the mitochondrial dynamics (fission/fusion) of the CLEC cells, which showed enhanced mitochondrial fragmentation and turnover rates. We also identified novel cell compensatory mechanisms that may mask the true extent of chronic Cd exposure induced damage in liver cells. CLEC and glucose work additively to damage hepatocellular mitochondrial function. New approach methodologies (NAMs), such as the current vitro toxicology study, establish the insidious effects of chronic heavy metal pollutant exposures on human hepatocellular function.

pharmacology and toxicology↗

bioRxiv2025

JointMap identified SERPINA3+ chondrocytes as a therapeutic target for osteoarthritis

BackgroundOsteoarthritis (OA) is a prevalent degenerative joint disease that affects nearly half of the population aged over 65. Recent single-cell transcriptional studies have identified heterogeneous OA-specific chondrocyte subtypes, yet a unified subtype-classification remains incomplete. MethodsIn this study, we integrated a total of 545,946 cells to develop JointMap, an interspecies atlas. The chondrocyte subtypes were analyzed using machine learning algorithms were validated via human samples, animal models, and in vitro cell culture. We performed KAS-seq and analyzed the spatial characters for JointMap. ResultsBased on clinical diagnosis and cross-omic indexes, we identified a novel OA-elevated chondrocyte subtype, SERPINA3+ chondrocytes. Notably, we uncovered their upregulated chondrogenesis function and intercellular communication patterns as potential therapeutic targets. ConclusionsIn summary, JointMap provides a unified framework for the systematic annotation of chondral tissues across species and helps in the identification of druggable networks for articular diseases.

bioinformatics↗

bioRxiv2025

Transcontinental Spread of HPAI H5N1 from South America to Antarctica via Avian Vectors

Our study has for the first time identified H5N1 strains (clade 2.3.4.4b, genotype B3.2) in brown skuas from the Fildes Peninsula, South Shetland Islands, Antarctica. These findings indicate that highly pathogenic avian influenza H5N1 viruses is now actively circulating in Antarctic ecosystems, representing a significant expansion of its geographic range.

genomics↗

bioRxiv2025

Asymmetric photosynthetic responses to hydrothermal variations between the two halves of the year in the Amazon rainforest

The Amazon rainforest, which stores approximately 120 billion tons of carbon and contributes around 16% of global terrestrial photosynthetic productivity, plays a pivotal role in global carbon cycling. Unlike temperate and boreal forests, tropical forests exhibit a bimodal photosynthetic pattern, characterized by distinct peaks in the first and second halves of the year. However, the intra-annual differences in photosynthetic responses to hydrothermal variations between these two periods in the Amazon rainforest remain largely unexplored. Here, utilizing satellite-derived photosynthetic proxies alongside ground-based flux tower observations from 2001 to 2020, we investigated the differences in photosynthetic responses to hydrothermal variations between the first and second halves of the year in the Amazon rainforest. Our observations revealed weaker temperature limitations but stronger precipitation limitations on photosynthesis in the second half of the year compared to the first half. Temperature constraints on photosynthesis have progressively weakened in both periods, while precipitation limitations have intensified, particularly in the latter half. Although the optimal temperature for photosynthesis is higher in the second half of the year, it is reached earlier, resulting in a sharper decline in photosynthetic productivity over the past two decades. Our findings reveal a shift from temperature to precipitation limitation in the Amazon, underscoring intra-annual asymmetry in vulnerability to intensifying heatwaves and droughts and calling for its explicit integration into management strategies and predictive models.

ecology↗

bioRxiv2025

Single-cell transcriptomics and machine learning reveal RNF144B and C5AR1 as immune-related biomarkers and therapeutic targets in myocardial infarction

BackgroundMyocardial infarction (MI) is a life-threatening cardiovascular disease characterized by high morbidity and mortality. Although advances in clinical management have improved patient outcomes, early diagnosis and effective immunomodulatory therapies remain limited. MethodsIn this study, we integrated multiple transcriptomic datasets and applied machine learning approaches, including LASSO regression, to identify a robust 13 key genes significantly associated with MI. Gene Set Enrichment Analysis (GSEA) and Gene Set Variation Analysis (GSVA) were subsequently performed to explore their potential biological functions. The immunological relevance of these genes was evaluated by analyzing their correlations with inflammation-related genes and those involved in immune cell migration. In addition, transcription factor (Johnson, Law et al.) and microRNA (miRNA) regulatory networks were constructed to elucidate upstream regulatory mechanisms. The expression levels of the 13 key genes were validated in MI mouse model. Furthermore, molecular docking was performed to identify candidate small molecules targeting core genes. ResultsAmong 11 cardiac cell populations identified, myeloid cells contributed most prominently to MI pathogenesis. A robust 13-gene predictive signature was established, with RNF144B and C5AR1 showing strong associations with immune modulation and disease severity. GSEA and GSVA further revealed that RNF144B was enriched in the neutrophil degranulation pathway, while C5AR1 was associated with the complement cascade. Correlation analysis demonstrated a significant positive relationship of RNF144B and C5AR1 with immunological roles. Both genes were also positively correlated with classical MI marker genes SERPINE1 and RUNX1. TF-gene and miRNA-mRNA regulatory networks supported the post-transcriptional regulation of these genes. In the MI mouse model, expression of the 13 genes was consistent with the risk-prediction model. Molecular docking identified CCX168 as a promising small-molecule candidate targeting RNF144B and C5AR1. ConclusionThis study reveals immune-related transcriptional signatures and signaling pathways that drive MI progression. The identified 13-gene signature, particularly RNF144B and C5AR1, holds promise as a diagnostic biomarker and therapeutic target, providing new insights for immunomodulatory and precision medicine strategies in MI. Keywords: Immune microenvironment, myeloid cells, inflammation, RNF144B, MI, immune-targeted therapy HighlightsO_LIA 13-gene signature was identified to predict MI risk C_LIO_LIRNF144B and C5AR1 are key immune regulators in MI progression C_LIO_LIGSEA and GSVA reveal immune-related pathways of RNF144B and C5AR1 C_LIO_LICCX168 is identified as a potential small-molecule therapy for MI C_LIO_LIMouse model validates gene expression consistent with the prediction model C_LI

cell biology↗

bioRxiv2025

Guess till correct: Gungnir codec enabling high error-tolerance and low-redundancy DNA storage through substantial computing power

DNA has emerged as a compelling archival storage medium, offering unprecedented information density and millennia-scale durability. Despite its promise, DNA-based data storage faces critical challenges due to error-prone processes during DNA synthesis, storage, and sequencing. In this study, we introduce Gungnir, a codec system using the proof-of-work idea to address substitution, insertion, and deletion errors in a sequence. With a hash signature for each data fragment, Gungnir corrects the errors by testing the educated guesses until the hash signature is matched. For practicality, especially when sequenced with nanopore long-read, Gungnir also considers biochemical constraints including GC-content, homopolymers, and error-prone motifs during encoding. In silico benchmarking demonstrates its outperforming error resilience capacity against the state-of-art methods and achieving complete binary data recovery from a single sequence copy containing 20% erroneous bases. Gungnir requires neither keeping many redundant sequence copies to address storage degradation, nor high-coverage sequencing to address sequencing error, reducing the overall cost of using DNA for storage.

bioinformatics↗

bioRxiv2025

Combining Selenocystine Based Fluorescence Assay with Structural Prediction for Enhanced Detection and Analysis of Cystine Stone Pathology

BackgroundCystine stones, a rare but recurrent type of kidney stones, primarily result from cystinuria, an inherited disorder caused by mutations in the genes SLC3A1 and SLC7A9, which encode renal cystine transporters. These mutations impair cystine reabsorption, resulting in elevated urinary cystine concentrations and stone formation. Current diagnostic methods are limited, particularly for detecting molecular-level dysfunctions. We aimed to develop a nonradioactive method for functional assessment of cystine transporters and mutation-specific pathologies. MethodsWe designed an innovative diagnostic approach combining a selenocystine-based fluorescence assay with structural predictions using AlphaFold to rapidly and accurately assess cystine transporter function and the molecular impacts of genetic mutations. ResultsOur assay demonstrated comparable transport efficiencies of cystine and selenocystine by the SLC3A1/SLC7A9 complex, and effectively differentiated mild, moderate, and severe functional impairments associated with known clinical mutations, including A354T and P482L. Structural modeling further provided mechanistic insights into mutation-induced dysfunctions. ConclusionThis integrated approach--combining a sensitive selenocystine fluorescence assay with AI-powered structural analysis--enables rapid, precise diagnosis of cystinuria variants and delivers mechanistic insights for personalized therapeutic strategies.

genetics↗

bioRxiv2025

CMDdemux: an efficient single cell demultiplexing method

Multiplexing technologies label cells with molecular tags, allowing cells from different donors to be pooled together for sequencing. Although this approach enhances cell throughput, eliminates batch effects, and enables doublet detection, limitations of hashtag-based labelling can still lead to low-quality data. Existing demultiplexing methods can accurately assign donor identities in high-quality datasets, but they often fail on low-quality data. To address this, we developed CMDdemux, a method comprising three key steps: within-cell centered log-ratio (CLR) normalization of hashtag count data, K-medoids clustering, and classification of cells based on Mahalanobis distance. By integrating both hashing and mRNA data, CMDdemux achieves high accuracy in distinguishing singlets, doublets, and negatives. It also provides visualization tools to help users inspect potentially misclassified droplets. We benchmarked CMDdemux against existing methods using a range of high- and low-quality datasets. Results show that CMDdemux consistently outperforms other approaches, demonstrating robust performance on both high- and low-quality data where other methods fail. CMDdemux is particularly effective in handling diverse types of low-quality multiplexing data across different multiplexing technologies.

bioinformatics↗

bioRxiv2025

An electrostatic repulsion model of centromere organisation

During cell division, chromosomes reorganise into compact bodies in which centromeres localise precisely at the chromatin surface1-4 to enable kinetochore-microtubule interactions essential for genome segregation5-8. The physical principles guiding this centromere positioning remain unknown. Here, we reveal that human core centromeres are directed to the chromatin surface by repulsion of centromere-associated proteins - independent of condensin-mediated loop extrusion and microtubule engagement. Using cellular perturbations, biochemical reconstitution, and multiscale molecular dynamics simulations, we show that chromatin surface localisation emerges from repulsion between condensed chromatin and both the kinetochore and the highly negatively charged centromere protein, CENP-B. Together, these elements form a centromeric region composed of two domains with opposing affinities, one favouring integration within the mitotic chromosome and the other favouring exposure to the surrounding cytoplasm, thereby driving surface positioning. Tethering synthetic negatively charged proteins to chromatin was sufficient to recapitulate this surface localisation in cells and in vitro, indicating that electrostatic repulsion is a key determinant of surface localisation. These findings demonstrate that centromere layering is not hardwired by chromatin folding patterns but instead emerges from phase separation in chromatin. Our work uncovers electrostatic polarity as a general and programmable mechanism to spatially organise chromatin.

cell biology↗

bioRxiv2025

Unveiling Eukaryotic Membrane Proteins in High Resolution Using Peptide Solubilization

Integral membrane proteins are vital for numerous biological functions and are typically studied using X-ray crystallography and cryo-electron microscopy (cryo-EM). However, these techniques require the extraction of target membrane proteins from their native membranes using detergents, which might disrupt the lipid environments and alter protein behavior. In this study, we present a novel method for solubilizing membrane proteins using a peptide, thereby eliminating the need for detergents throughout the procedure. We demonstrate that the 4F peptide effectively solubilizes a range of membrane proteins and complexes into peptidiscs, while preserving their functionality and structural integrity. Converting these peptidiscs into nanodiscs further enhances particle homogeneity and facilitates high-resolution structural determination of membrane proteins. Our findings highlight the potential of membrane-solubilizing peptides to advance membrane protein research. HighlightsO_LI4F peptide can effectively solubilize eukaryotic membrane proteins C_LIO_LI4F-peptidisc can be converted to MSP-wrapped nanodisc for high resolution cryo-EM studies C_LIO_LI4F solubilization can preserve endogenous ligands bound to the target membrane protein C_LI

biophysics↗

bioRxiv2025

Processing bodies promote lysosomal quality control and cell survival during recovery from lysosomal damage

Lysosomes are essential for cell survival but are highly susceptible to diverse physical and pathological stressors. Thus, the ability to initiate an acute damage response and promote recovery after stressor resolution is critical for maintaining cellular homeostasis and viability. Although recent studies have advanced our understanding of acute responses to lysosomal injury, the molecular mechanisms governing the recovery stage and distinguishing it from the acute phase remain poorly defined. Here, we delineate a key difference between these two stages in translational regulation and uncover lysosomal recovery from acute damage as a novel trigger for processing body (PB) formation. PBs are membraneless biomolecular condensates involved in RNA metabolism and translational reprogramming. We provide the first evidence that PBs are critical for lysosomal quality control and cell survival during recovery. Mechanistically, PBs are induced selectively during the recovery phase, but not during the acute damage response, through interactions with stress granules (SGs), distinct membraneless biomolecular condensates formed upon acute injury to stabilize damaged lysosomal membranes for repair. Functional analyses reveal that PBs promote lysosomal quality control by collaborating with SG-mediated membrane stabilization, while independently recruiting released cathepsins, thereby collectively supporting cell survival. Together, these findings establish PBs as central effectors of the lysosomal recovery program and underscore the broader relevance of biomolecular condensates in cellular responses to lysosomal damage and related disease processes.

cell biology↗

bioRxiv2025

STARNet enables spatially resolved inference of gene regulatory networks from spatial multi-omics data

Biological tissues are composed of distinct microenvironments that spatially orchestrate gene expression and cell identity. However, the regulatory principles governing domain-specific cellular functions remain poorly understood due to the lack of effective methods for mapping gene regulatory networks (GRNs) in situ. To address this gap, we introduce STARNet, a representation learning approach that leverages heterogeneous hypergraph modeling of spatial transcriptomic and epigenomic data to resolve tissue-domain-specific regulatory interactions. By integrating graph neural networks with contrastive learning in a self-supervised framework, STARNet learns unified embeddings that preserve both multi-modal molecular features and anatomical spatial context, enabling accurate and domain-resolved GRN reconstruction within complex tissues. Benchmarking on both simulated and real datasets demonstrates that STARNet achieves state-of-the-art performance. We further demonstrate its broad applicability across diverse biological contexts, including neural development, genetic disease risk, and drug-induced developmental toxicity. In the mouse brain, it delineates region-specific regulatory networks and reconstructs spatiotemporal programs underlying neural stem cell differentiation. In human genetics, it provides a mechanistic link between genotypes and phenotypes by showing how genome-wide association study (GWAS) variants for complex diseases perturb hippocampus-specific GRNs. In developmental toxicology, STARNet reveals that drug-induced disruptions of GRNs in defined embryonic regions underlie tissue-specific vulnerability. Collectively, STARNet offers a powerful and versatile framework for resolving the spatial regulatory logic of complex tissues, providing multi-angle insights into tissue patterning, development, and disease mechanisms.

bioinformatics↗

bioRxiv2025

Systematic Evaluation of Somatic Contamination in Germline Genomes

Large-scale genomic initiatives like the UK Biobank (UKB) have revolutionized our understanding of human disease. These studies typically assume that blood-derived DNA faithfully reflects an individuals germline genome. However, this assumption is challenged by somatic mutations arising from processes like clonal hematopoiesis. Although standard bioinformatics pipelines employ variant allele frequency (VAF)-based filtering to mitigate such contamination, the efficacy of these approaches requires systematic evaluation. By systematically analyzing whole-exome sequencing (WES) and whole-genome sequencing (WGS) data from large cohorts, including the UK Biobank, The Cancer Genome Atlas (TCGA), and the 1000 Genomes Project, we revealed critical limitations in current filtering methodologies. We found that the mutational spectrum of rare "germline" variants is highly similar to that of somatic mutations. Furthermore, we uncovered these variants show significant associations with phenotypes such as age, sex, and smoking status, established drivers of somatic mutagenesis. This persistent somatic contamination introduces substantial confounder effects, potentially generating spurious associations and reverse causality in genetic studies. Our work underscores the urgent reconsideration of two fundamental aspects of genomic research: (1) refinement of variant filtering strategies to better distinguish true germline variants from somatic contaminants, and (2) incorporation of somatic mutagenesis factors as essential covariates in study design. Our findings provide a basic framework for improving the accuracy and interpretability of large-scale genomic studies.

genomics↗

bioRxiv2025

High-throughput 3D super-resolution ultrasound imaging

Capturing fast hemodynamics in deep organs is essential for understanding microvascular regulation of organ responses to physiological demands and pathological stress. However, non-invasive three-dimensional (3D) imaging of these microscale processes remains challenging due to trade-offs between spatial resolution, imaging speed, and penetration depth. Here, we present fluctuation-based high-order super-resolution acoustic microscope (FLAME), a tracking-free 3D ultrasound imaging technology capable of fast microvascular angiography and flow measurement. Using as few as 30 volumes, FLAME improves 3D ultrasound resolution by 8-fold ([~]50 m) and shortens data acquisition time by over two orders of magnitude, from tens of seconds to tens of milliseconds, compared with conventional tracking-based super-resolution approaches1-6. This high-throughput capability supports high-fidelity imaging of tissue functions under flexible experimental settings. FLAME achieves an unprecedented 3D super-resolution frame rate of [~]40 Hz, capturing transient hemodynamic responses to various physiological and pathological stimuli in different mouse organs, from the microvascular level to the whole-body scale. With open-sourced implementation, FLAME provides an accessible platform for real-time, in-depth hemodynamic monitoring across diverse biomedical applications and clinical translation.

bioengineering↗

bioRxiv2025

Multi-omics reveals lipid metabolism, mitochondrial and extracellular matrix dysregulation of diabetic cardiomyopathy in the human heart

Diabetic cardiomyopathy (DbCM) is a condition characterized by myocardial dysfunction in diabetes. Although the clinical recognition of DbCM is increasing, its underlying molecular mechanisms remain poorly understood. We aimed to conduct a comprehensive multi-omics analysis of heart tissue from cardiomyopathy patients (both with and without diabetes),alongside tissue from healthy donors, by integrating transcriptomic, 4D-DIA proteomic, and full-spectrum widely targeted metabolomic data. Differential analyses revealed a significant metabolic pattern in patients with DbCM, characterized by increasing utilization of triglyceride-derived fatty acids. Correlation analyses highlighted impaired BNIP3L-mediated mitophagy as a potential contributor to the disruption of fatty acid metabolism and mitochondrial dysfunction in DbCM. Protein interaction and fibrotic trichrome analyses showed that patients with DbCM had impaired cardiac matrix repair compared to patients with cardiomyopathy without diabetes. These findings provide a system-level understanding of DbCM and reveal molecular signatures driving its progression, offering insights for targeted therapeutic strategies against diabetic heart disease.

bioinformatics↗