Biology subjects

Wang, B.

Publications and source records attributed to Wang, B..

At least 181 records · Page 10Linked to original sources

bioRxiv2025

A genome-wide MAGIC kit for recombinase-independent mosaic analysis in Drosophila

Mosaic analysis has been instrumental in advancing developmental and cell biology. Most current mosaic techniques rely on exogenous site-specific recombination sequences that need to be introduced into the genome, limiting their application. Mosaic analysis by gRNA-induced crossing-over (MAGIC) was recently developed in Drosophila to eliminate this requirement by inducing somatic recombination through CRISPR/Cas9-generated DNA double-strand breaks. However, MAGIC has not been widely adopted because gRNA-markers, a required component for this technique, are not yet available for most chromosomes. Here, we present a complete, genome-wide gRNA-marker kit that incorporates optimized designs for enhanced clone induction and more effective clone labeling in both positive MAGIC (pMAGIC) and negative MAGIC (nMAGIC). With this kit, we demonstrate clonal analysis in a broad range of Drosophila tissues, including cell types that have been difficult to analyze using recombinase-based systems. Notably, MAGIC enables clonal analysis of pericentromeric genes and deficiency chromosomes and in interspecific hybrid animals, opening new avenues for gene function study, rapid gene discovery, and understanding cellular basis of speciation. This MAGIC kit complements existing systems and makes mosaic analysis accessible to address a wider range of biological questions. IMPACT STATEMENTA comprehensive toolkit enables genome-wide, recombinase-independent mosaic analysis in Drosophila, permitting clonal analysis of pericentromeric genes, deficiency chromosomes, and interspecific hybrids previously inaccessible to standard methods.

genetics↗

bioRxiv2025

Pan-cancer prediction of tumor immune activation and response to immune checkpoint blockade from tumor transcriptomics and histopathology

Accurately predicting which patients will respond to immune checkpoint blockade (ICB) remains a major challenge. Here, we present TIME_ACT, an unsupervised 66-gene transcriptomic signature of tumor immune activation derived from TCGA (The Cancer Genome Atlas) melanoma data. First, we demonstrate that TIME_ACT scores accurately identify tumors with activated immune microenvironments across different cancer types. Further, analysis of spatial features reveals that tumor microenvironment regions with dense lymphocyte infiltration near tumor cells have high TIME_ACT scores, successfully marking localized immune activation. Second, across 25 transcriptomic ICB cohorts encompassing nine cancer types, TIME_ACT achieves a mean AUC of 0.76 and a mean odds ratio of 5.77, significantly outperforming 30 established transcriptomic signatures and prediction methods for ICB response, including a recently developed foundation model for immunotherapy response prediction. Third, we show that TIME_ACT scores can be accurately inferred from routine tumor histopathology slides and that slide-inferred TIME_ACT scores predict ICB response across nine new independent patient cohorts spanning eight cancer types, achieving a mean AUC of 0.72 and a mean odds ratio of 4.99. These findings establish TIME_ACT as a robust, pan-cancer biomarker that enables accurate, low-cost, and clinically scalable prediction of ICB response from routine histopathology.

cancer biology↗

bioRxiv2025

Interpreting Attention Mechanisms in Genomic Transformer Models: A Framework for Biological Insights

Transformer models have shown strong performance on biological sequence prediction tasks, but the interpretability of their internal mechanisms remains underexplored. Given their application in biomedical research, understanding the mechanisms behind these models predictions is crucial for their widespread adoption. We introduce a method to interpret attention heads in genomic transformers by correlating per-token attention scores with curated biological annotations, and we use GPT-4 to summarize each heads focus. Applying this to DNABERT, Nucleotide Transformer, and scGPT, we find that attention heads learn biologically meaningful associations during self-supervised pre-training and that these associations shift with fine-tuning. We show that interpretability varies with tokenization scheme, and that context-dependence plays a key role in head behaviour. Through ablation, we demonstrate that heads strongly associated with biological features are more important for task performance than uninformative heads in the same layers. In DNABERT trained for TATA promoter prediction, we observe heads with positive and negative associations reflecting positive and negative learning dynamics. Our results offer a framework to trace how biological features are learned from random initialization to pre-training to fine-tuning, enabling insight into how genomic foundation models represent nucleotides, genes, and cells.

bioinformatics↗

bioRxiv2025

Biological Reasoning with Reinforcement Learning through Natural Language Enables Generalizable Zero-Shot Cell Type Annotations

Single-cell RNA-sequencing (scRNAseq) has reshaped biomedical research, enabling the high-resolution characterization of cellular populations. Yet cell type annotation, a process typically performed by domain experts interpreting gene expression patterns by manual curation or with specialized algorithms, remains labor-intensive and limited by prior knowledge. In addition, while reasoning large language models (LLMs) have demonstrated remarkable performance on mathematics, coding and general-reasoning benchmarks, their potential in scRNAseq analyses remains underexplored. Here, we investigate the advantages and limitations of employing DeepSeek-R1-0528, a recently developed open-source 671B-parameter reasoning LLM, for zero-shot scRNAseq cell type annotation. We find that DeepSeek-R1 prompted with a ranked list of 10 differentially expressed marker genes per cluster of single cells outperforms both its reasoning-enhanced, non-reasoning equivalent (DeepSeek-V3-0324) and GPT-4o in cluster-level annotations. At the level of single cells, DeepSeek-R1 prompted with the top 500 expressed genes in a cell outperforms its non-reasoning counterpart DeepSeek-V3, illustrating test-time scaling for bioinformatics tasks through natural language. Running DeepSeek-R1 in zero-shot classifier mode, with a prompt that presents a broad catalogue of cell type labels to choose from, improves its performance and generalizability across different datasets. On data curated by the expert model scTab (termed in-domain data), the DeepSeek-R1 classifiers perform better than the expert model scGPT and on par with the specialized cell genomics LLM C2S-Scale-1B, but lag behind scTab. On out-of-distribution data unseen by the two expert models, DeepSeek-R1 and its classifier versions generalize better and outperform the other models in the majority of the evaluated datasets. Notably, DeepSeek-R1 supports its cell type calls with interpretable textual biological rationales underlying its reasoning, providing a learning opportunity for researchers. Nevertheless, peak annotation performance remains modest, highlighting the intrinsic complexity of scRNAseq cell type annotation.

bioinformatics↗

bioRxiv2025

Pre-search attentional focus supports learned suppression in visual search

Salient distractors often capture our attention, disrupting ongoing tasks. Recent studies suggest that, through statistical learning, prior experiences regarding distractor locations can reduce distraction by suppressing their corresponding locations. However, the proactive neural mechanisms supporting this learned suppression remain unclear. Our findings demonstrate that participants learn to suppress locations that are more likely to contain distractors relative to other locations. Using frequency tagging in electroencephalography (EEG) recordings, we observed significantly different tagging responses between high- and low-probability locations, along with a general decrease in alpha power (8-12 Hz) prior to search onset. Notably, the higher tagging frequency power at high-probability locations suggests that participants allocated greater attentional focus to these locations in anticipation of the search. These results suggest that anticipatory attentional deployment precedes the suppression of high-probability distractor locations after the onset of visual search.

neuroscience↗

bioRxiv2025

Catalytic pocket-informed augmentation of enzyme kinetic parameters prediction via hierarchical graph learning

Predicting enzyme kinetic parameters is crucial for enzyme engineering and mining, but development of accurate tools for this task is still challenging due to the complexity of influencing factors. Large language models (LLMs)-based deep learning methods show satisfactory performance in Kcat prediction by encoding substrate and enzyme sequence information, but this further enhancement of the performance can be largely limited by the ignorance of the substrate-binding information. Here, we introduce GraphKcat, a deep learning framework that integrates enzyme-substrate 3D binding conformations for precise kinetic parameter prediction. The method employs Chai-1 to generate enzyme-substrate complex conformations, followed by a developed multi-scale hierarchical graph neural network to systematically characterize active site features from all-atom (AA) level to coarse-grained (CG) level, which are subsequently fused with LLMs embedded substrate, sequence, and environmental related information for prediction. To enable efficient multimodal feature integration, we proposed a multi modal cross-attention fusion (MMCAF) module for feature alignment and update. Experimental evaluations demonstrate that GraphKcat can effectively identifying catalysis-critical residues, learning sequence similarity-independent features through its multi-scale graph architecture, and maintaining robust performance under low sequence similarity conditions. Critically, GraphKcat captures conserved structural and physicochemical patterns related to enzymatic activity, enabling accurate identification of high-activity enzyme variants and functional mutants even for low-homology targets. These capabilities highlight its potential as a transformative tool for enzymatic activity prediction, rational enzyme engineering, and enzyme mining in industrial applications.

bioinformatics↗

bioRxiv2025

Macroscopic Analyses of RNA-Seq Data to Reveal Chromatin Modifications in Aging and Disease

Regulation of gene expression is fundamental for proper cellular function, and is constrained by the local chromatin environment of each gene, which varies spatially along the chromosome and is shaped by epigenetic modifications. Epigenetic modifications induce changes in the local chromatin structure, which can influence gene expression, by affecting the accessibility of DNA to transcription factors. Such changes are particularly relevant in aging and genetic disorders like Hutchinson-Gilford Progeria Syndrome (HGPS) and Werner Syndrome (WRN), where altered chromatin structure contributes to disease pathology. In this study, we analyze RNA-seq data using macroscopic metrics designed to be explicitly sensitive to chromatin modifications. The first metric, intra-chromosomal gene correlation length, measures spatial correlations in gene expressions along the chromosome. The second metric employs an energy landscape model based on the Arrhenius equation to estimate the energetic barriers associated with chromatin state transitions. We apply these metrics to various aging-related datasets, demonstrating their sensitivity to changes in the chromatin structure and the interpretability of the resulting outputs. The intra-chromosomal gene correlation length is particularly effective in quantifying changes in RNA-seq profiles due to increased chromatin accessibility during aging (and conversely, reduced accessibility due to treatment). This metric not only accurately distinguishes cell states, but also provides insight into the direction of aging. For instance, our observations on the effects of anti-sense oligonucleotide (ASO) treatment align with the existing literature, demonstrating that ASO partially restores chromatin structure in diseased cells. They additionally quantify the more pronounced effects in HGPS compared to WRN. The barrier energy landscape further extends this capability by offering a framework for understanding the progressive degradation of the regulatory mechanisms. Together, these metrics provide robust screening tools that enhance our ability to exploit common measurements such as RNA-seq to derive new phenotypes such as chromatin dynamics on aging and disease, offering an alternative perspective that complements traditional analytical techniques and enriches our understanding of cellular states.

bioinformatics↗

bioRxiv2025

Humans underestimate their body mass in microgravity: evidence from reaching movements during spaceflight

Astronauts consistently exhibit slower movements in microgravity, even during tasks requiring rapid responses. The sensorimotor mechanisms underlying this general slowing remain debated. Two hypotheses have been proposed: either the sensorimotor system adopts a conservative control strategy for safety and postural stability, or the system underestimates body mass due to reduced inputs from proprioceptive receptors. To dissociate these opinions, we studied twelve taikonauts aboard the China Space Station performing a classical hand-reaching task. Compared to their pre-flight performance and to an age-matched control group, participants showed increased movement durations and altered kinematic profiles in microgravity. Model-based analyses of motor control parameters revealed that these changes stemmed from reduced initial force generation in the feedforward control phase followed by compensatory feedback-based corrections. These findings providing support for the body mass underestimation hypothesis while being inconsistent with the strategic slowing hypothesis. Importantly, the sensory estimate of bodily property in microgravity is biased but immune from sensorimotor adaptation, calling for an extension of existing theories of motor learning.

neuroscience↗

bioRxiv2025

Statistical learning drives anticipatory micro-saccades toward suppressed distractor locations

Statistical learning enables individuals to suppress locations associated with salient distractors, yet the mechanisms underlying this suppression remain unclear. Proactive accounts propose that suppression operates without prior attentional allocation to distractor locations, whereas reactive accounts suggest that suppression requires covert attention to the distractor location before it can be engaged. To address this ongoing debate, the current study recorded micro-saccades--an index of covert attention--during the pre-stimulus interval, alongside electroencephalogram (EEG) data collected during a visual search task. Participants were instructed to ignore a salient distractor that occurred more frequently at a specific location. Statistical learning reduced attentional capture for distractors presented at this high-probability location, and this effect was accompanied by oculomotor markers: micro-saccade rates decreased before stimulus onset relative to a control condition. Strikingly, these anticipatory micro-saccades were more often directed toward high-probability distractor locations than away from them, consistent with a reactive suppression mechanism. In parallel, alpha-band activity (8-14 Hz) carried decodable representations of high-probability distractor locations, indicating preparatory neural tuning. Overall, these findings provide evidence that the oculomotor system is closely involved in encoding and responding to learned spatial regularities.

neuroscience↗

bioRxiv2025

Structural Basis for Retron Co-option of Anti-phage ATPase-nuclease

Retrons are recently identified bacterial defense systems that employ a tripartite of reverse transcriptase, non-coding RNA (ncRNA) and its derived multi-copy single stranded DNA (msDNA) to sequester effector activity. Phage invasion activates retrons, triggering effector activity and inducing abortive infection and cell growth arrest. Ec78 distinguishes itself from other retrons by leveraging the Septu defense system, a stand-alone ATPase-nuclease pair (PtuAB), by reshaping its phage sensing and molecular assembly processes. To elucidate the evolutionary adaptation of PtuAB by Ec78, we determined electron cryomicroscopy structures of Ec78 as well as the retron-displaced PtuAB at near atomic resolutions. We show that the Ec78-associated ATPase, PtuA, acquired unique elements that enable its interactions with the reverse transcriptase and the msDNA, as well as self-assembly when displaced by the retron. Supported by biochemical and mutational analysis, we also show that the retron-displaced PtuAB forms a tetramer, unlike its stand-alone counterpart, that indiscriminately restricts the host but, in the presence of the retron, confers a well-controlled immune response, eliciting ATP hydrolysis- and msDNA-regulated targeting to host tRNATyr. Our studies reveal an evolutionary principle for retrons to co-opt conserved enzyme modules for defense in response to different cellular needs. HighlightsO_LIEc78 reverse transcriptase and msDNA sequester PtuAB toxicity C_LIO_LIEc78 PtuAB heterocomplex targets host tRNATyr C_LIO_LIATPase and HNH activities are required for Ec78 PtuAB function C_LIO_LIEc78 PtuA evolved novel domains for binding retron and for self-assembly C_LIO_LIEc78 requires distinct ncRNA and msDNA processing for binding PtuAB C_LIO_LIEc78 retron facilities efficient human genome editing C_LI

biophysics↗

bioRxiv2025

Intracellular Delivery of Bone Nanoparticles to Mitigate Irradiation-Induced Damage in Bone Marrow Mesenchymal Stem Cells

Ionizing radiation (IR)-induced bone damage presents a major clinical challenge by impairing bone marrow function and disrupting normal bone remodeling. Bone regeneration depends on bone marrow-derived mesenchymal stem cells (BMSCs), which are highly sensitive to IR that causes DNA damage, oxidative stress, apoptosis, and a shift from osteogenesis to adipogenesis, ultimately leading to bone loss and impaired healing. This study evaluated the therapeutic potential of intracellularly delivered bone-derived nanoparticles (BPs) in mitigating IR-induced BMSCs damage. We found that IR exposure caused significant BMSCs dysfunction, including reduced proliferation, increased apoptosis, persistent DNA damage, and a shift toward adipogenic differentiation. Treatment with BPs led to efficient intracellular uptake, improved cell morphology, enhanced proliferation, reduced apoptosis, and preservation of balanced differentiation capacity. Transcriptomic analysis via RNA sequencing revealed that BPs restored key molecular pathways disrupted by IR, particularly those involved in cell cycle regulation, extracellular matrix (ECM) remodeling, and apoptosis. By reversing these transcriptional impairments, BPs supported genomic stability and the regenerat ive function of BMSCs. Overall, these findings suggest that BPs effectively counteract IR-induced cellular damage and enhance the regenerative capacity of BMSCs, offering a promising therapeutic strategy for radiation -induced skeletal injuries.

bioengineering↗

bioRxiv2025

Adolescent intermittent ethanol exposure induces sex-specific and time-dependent changes in affective behaviors and metabolomic profiles.

Given the adverse effects of adolescent binge alcohol consumption in humans, the present work explores the lasting behavioral and metabolomic impacts following adolescent binge ethanol exposure. Here we determine short- and long-term changes in affective behaviors and metabolomic profiles in male and female adolescent binge ethanol-exposed mice. Male and female C57BL/6J mice were exposed to adolescent intermittent ethanol (AIE) as a model of binge ethanol exposure using intermittent vapor inhalation from postnatal day (PND) 28-42. Mice were tested in a battery of behavioral tests and metabolomic analyses were conducted following short-term and long-term withdrawal from AIE exposure. Mice were tested for affective behaviors using the open field test (OFT), the light/dark test (LDT), and the tail suspension test (TST) one week following AIE exposure from PND 49-53 and again from PND 91-95. Serum samples were collected on PND 43, corresponding to 24 hours after the last exposure, fecal samples were collected during each OFT, and liver samples were collected approximately 80 days after AIE exposure for metabolomic analysis. We show modest incubation of behavioral differences in anxiety-like behavior in males after adolescent binge ethanol exposure; an effect that was absent in female mice. In contrast, metabolomic differences in male mice that were more pronounced shortly after adolescent binge ethanol exposure waned as time progressed since last exposure. Male mice appear to be more susceptible to the persistent changes in adolescent binge ethanol exposure that varies over time. It is possible that short-term metabolomic changes may predict long-term behavioral changes in affective behaviors. Contribution to the field statementAdolescent binge alcohol exposure causes long-lasting changes in behavior, however the underlying biology mediating these changes have not readily been assessed in a sex-specific manner. Metabolomic assays are a powerful tool that can be used as biomarkers to determine changes in biological pathways that are linked to specific behavioral phenotypes. Here we explored the relationship between adolescent binge alcohol exposure, changes in anxiety-like behavior, and metabolomic profiles in male and female C57BL/6J mice. We find robust changes in metabolomic profiles following short-term withdrawal in male mice exposed to binge alcohol during adolescence. We also see modest changes in anxiety-like behavior following long-term withdrawal. Although female mice did not show any robust changes in anxiety-like behavior nor global changes in metabolomic profiles, in both sexes we do observe persistent changes in some amino acids, which may serve as specific biomarkers associated with changes in alcohol-induced behavioral changes. These data add to the field by conducting a novel longitudinal study following male and female mice from adolescence to early adulthood and measuring physiological biomarkers (metabolomics) coupled with behavioral changes following binge ethanol exposure.

neuroscience↗

bioRxiv2025

Auranofin Inhibits Hepatitis E Virus Replication Via Reactive Oxygen Species

Hepatitis E virus (HEV) causes roughly 20 million yearly global infections, and is associated with chronic hepatitis, neurological sequelae and pregnancy-related adverse outcomes that require antiviral therapeutic intervention. While there are currently no approved HEV-specific therapeutics, ribavirin and pegylated interferon, prescribed off-label, are the current standard of care. However, ribavirin resistance and toxicity highlight the unmet clinical need to identify safer, HEV-specific antivirals. Auranofin, an FDA-approved anti-rheumatic drug, displays antiviral activity against several viruses. Therefore, we investigated auranofins potential as an antiviral and its mechanism of action against HEV. We demonstrated that auranofin displays dose-dependent antiviral activity against two genotypes of HEV that cause a significant proportion of human disease, as well as against a ribavirin treatment failure-associated mutant. Because auranofin is known to increase reactive oxygen species (ROS), we investigated the antiviral mechanism of action via treatment with ROS inhibitors. ROS inhibitors reversed auranofin-mediated ROS promotion and antiviral activity, suggesting the observed antiviral effects are mediated by ROS. Furthermore, treatment with a different ROS promotor, D-amino acid oxidase (DAAO), also displays antiviral activity against HEV, which was also reversed by treatment with a ROS inhibitor, suggesting that ROS accumulation alone is antiviral. We also demonstrated that combined treatment with auranofin and ribavirin exhibits synergistic antiviral activity in vitro, which supports repurposing auranofin as an antiviral against HEV, potentially in combination with ribavirin. Overall, this study has important implications in repurposing auranofin as an antiviral against HEV and in delineating the mechanism of action against HEV via ROS. ImportanceHepatitis E virus (HEV) lacks approved antiviral therapies, and off-label treatments are limited by toxicity and emerging resistance. This study identifies the FDA-approved drug auranofin as an effective in vitro inhibitor of HEV, including two globally relevant human-associated genotypes and a ribavirin treatment failure-associated mutant. Auranofins activity highlights the therapeutic potential of host-targeting antivirals, particularly those that promote the generation of reactive oxygen species, in treating HEV infection. These findings support further in vivo investigations of auranofin as a treatment for HEV and suggest that modulating host redox pathways by promoting reactive oxygen species may represent a promising strategy for broad-spectrum antiviral development.

microbiology↗

bioRxiv2025

Label-Free In-Line Characterization of Immune Cell Culture using Quantitative Phase Imaging

Cell therapies, including T cell immunotherapies, offer promising treatments for previously untreatable diseases, but their widespread use is hindered by challenges in monitoring therapeutic cells during culture--impacting consistency, potency, and cost. This work demonstrates the use of quantitative phase imaging (QPI), specifically a compact, non-interferometric form called quantitative oblique back illumination microscopy (qOBM), for non-destructive, label-free, in-line assessment of T cell cultures. qOBM enables near real-time feedback on culture growth, contamination, and cell status (viability and activation), comparable to flow cytometry. We further apply this method to characterize genetically modified CAR T cells and explore its potential for advanced T cell phenotyping. Analysis of data from over 50 independent donors shows strong correlation between qOBM metrics and traditional destructive at-line assays. Overall, qOBM provides a powerful tool for continuous, in-line monitoring of therapeutic cell cultures, which can be transformative for improving reproducibility, reducing costs, and advancing the development of cell-based therapies.

immunology↗

bioRxiv2025

PCSK9 dependent cholesterol acquisition and utilization underlies metastatic organ preference in pancreatic cancer.

To grow at distant sites, metastatic cells must overcome major challenges posed by the unique cellular and metabolic composition of secondary organs1. Pancreatic ductal adenocarcinoma (PDAC) is an aggressive disease that metastasizes to the liver and lungs. Despite evidence of metabolic reprogramming away from the primary site, the key drivers that dictate the ability of PDAC cells to colonize the liver or lungs and survive there are undefined. We identified Proprotein Convertase Subtilisin/Kexin Type 9 (PCSK9) as predictive of colonization of the liver versus lungs by integrating datasets describing the metastatic tropism of human PDAC cell lines2, with in vivo metastasis modeling in mice and gene expression correlation analysis. PCSK9 is a negative regulator of low density lipoprotein(LDL)-cholesterol import and, accordingly, PCSK9-low PDAC cells exhibit strong preference for colonizing LDL-rich liver tissue. Cholesterol taken up by liveravid PCSK9-low cells is converted into the signaling oxysterol, 24(S)-hydroxycholesterol, which reprograms the surrounding microenvironment to induce nutrient release from neighboring hepatocytes. Conversely, PCSK9-high, lung-avid PDAC cells rely on transcriptional upregulation of the distal cholesterol synthesis pathway to generate intermediates, chiefly, 7dehydrodesmosterol and 7-dehydrocholesterol, with protective action against ferroptosis, a vulnerability in the oxygen-rich microenvironment of the lung. Increasing PCSK9 levels redirected liver-avid cells to the lung whereas its ablation drove lung-avid cells to the liver, thereby establishing PCSK9 as necessary and sufficient for secondary organ site preference. Our studies reveal PCSK9-driven differential utilization of the distal cholesterol synthesis pathway as a key and potentially actionable driver of metastatic growth in PDAC.

cancer biology↗

bioRxiv2025

Splicing factor SRSF1 is a pH-stat to restore nucleolar integrity and function

Nucleolus is a multiphase biomolecular assembly containing three distinct subcompartments. Evidence is just emerging that nucleolar proteins with characteristic electrochemical properties condense to generate a pH gradient, which serves as a key driving force to organize the nucleolar sub-phases. Given the indispensable functionality of nucleolus in most, if not all, of the cellular activities, it is vital for a cell to sense potential fluctuation in the nucleolar pH and subsequently maintain the pH homeostasis. The mechanisms how a cell achieve this task remains poorly decoded. Here, we show that splicing factor SRSF1 is shuttled from nuclear speckles (NSs) to the nucleolus in times of stress that interrogates the nucleolar pH. SRSF1 nucleolar localization is reliant on an acidic patch and molecular interactions with a nucleolar-resident protein, DDX18. Loss of SRSF1 impedes pH homeostasis in the nucleolus, and subsequently obstructs the restoration of the nucleolar multiphase and function. The arginine residues in the RS (arginine/serine rich) domain of SRSF1, endowed by high isoelectric point (pI), directly alkalize the nucleolar microenvironment. Interestingly, synthetic arginine-rich dipeptides derivative of SRSF1 RS domain safeguard nucleolus from pH and functional disturbance. Our findings uncover unprecedented mechanistic insights into nucleolar pH-sensing and regulation.

cell biology↗

bioRxiv2025

Bridging macroscopic and microscopic modeling of electric field by brain stimulation

BackgroundModeling the electric field (E-field) on the microscopic scale improves our understanding of brain stimulation modalities and modeling methods but requires careful consideration of the conductivity values and correction of the field amplitudes to match conventional models on the macroscopic scale. ObjectiveWe analyze the correction step and discuss its relevance and implications for E-field modeling efforts bridging the macroscopic and microscopic scales. MethodsWe provide the theoretical framework for comparing microscopic and macroscopic models and describe approaches for effectively and efficiently matching the E-field amplitude and tissue conductivity. ResultsConsistent results can be obtained for brain stimulation models on different scales with appropriately selected conductivity and E-field amplitudes. ConclusionMicroscopic E-field models enable numerical estimation of conductivity of macroscopic homogenous neural tissue from microscopically realistic brain samples and exploration of the effect of microscopic E-field perturbations on neural activation threshold by brain stimulation, therefore improving modeling accuracy.

bioengineering↗

bioRxiv2025

Transfer Learning and Permutation-Invariance improving Predicting Genome-wide, Cell-Specific and Directional Interventions Effects of Complex Systems

With the advent of precision medicine, single-drug treatments may not fully satisfy the pursuit of precision medicine. However, single-drug treatments have accumulated a large amount of data and a considerable number of deep learning models. In this context, using transfer learning to effectively leverage the existing vast amount of single-compound intervention effect data to build models that can accurately predict the intervention effects of complex systems is highly worth investigating. In this study, we used a deep model based on permutation-invariance as the core module, pre-trained on a large amount of single-compound intervention data in cell lines, and fine-tuned on a small amount of complex system (like natural products) intervention data in cell lines, resulting in a predictive model named SETComp (the Concat version with ~200M parameters and the Add version with ~173M parameters). The two versions of SETComp achieved an accuracy of 93.86% and 92.70%, respectively, on the complex system-cell-gene association test set, improving by 5.82% to 27.59% compared to the baseline. When predicting the intervention effects of those complex systems the model had never encountered before, the accuracy increased by up to 24.83% compared to the baseline. In our in vitro validation, up to 88.65% of the predictions were confirmed to be correct, and the models output showed a significant positive correlation with the real-world foldchange. We further observed SETComps potential in various biomedical scenarios, achieving good performance in applications such as mechanism uncovering, repositioning, and compound synergy discovery.

systems biology↗