Biology subjects

Guo, S.

Publications and source records attributed to Guo, S..

At least 55 records · Page 3Linked to original sources

bioRxiv2025

Evaluating scientific theories as predictive models in language neuroscience

Modern data-driven encoding models are highly effective at predicting brain responses to language stimuli. However, these models struggle to explain the underlying phenomena, i.e. what features of the stimulus drive the response? We present Question Answering encoding models, a method for converting qualitative theories of language selectivity into highly accurate, interpretable models of brain responses. QA encoding models annotate a language stimulus by using a large language model to answer yes-no questions corresponding to qualitative theories. A compact QA encoding model that uses only 35 questions outperforms existing baselines at predicting brain responses in both fMRI and ECoG data. The model weights also provide easily interpretable maps of language selectivity across cortex; these maps show quantitative agreement with meta-analyses of the existing literature and selectivity maps identified in a follow-up fMRI experiment. These results demonstrate that LLMs can bridge the widening gap between qualitative scientific theories and data-driven models.

neuroscience↗

bioRxiv2025

Walking Selectively Modulates Behaviorally Relevant Auditory Responses According to Stride Cycle Phase

Perception in natural environments unfolds while the body is in motion, yet how locomotion shapes cortical computations for fine-grained and dynamic scene analysis remains unclear. Here, we combined over-ground walking with mobile EEG and a transitional click-train paradigm to test how natural locomotion influences auditory processing in humans. Participants listened to two-segment click trains in which a subtle inter-click-interval (ICI) change signaled a pitch shift and performed a task-engaged change-detection task while either sitting or walking; an independent cohort also completed a passive-listening version. Walking elevated behavioral detection thresholds and slowed responses, indicating reduced sensitivity to fine temporal changes. Neurally, locomotion produced a response-component-specific attenuation: activity time-locked to the ICI transition--the change response previously linked to temporally integrative updating of an ongoing auditory object--was strongly reduced, whereas the offset response was largely preserved and onset responses were only modestly affected. Critically, this modulation persisted during passive listening and was selectively amplified near the perceptual threshold during task-engaged detection, revealing an interaction between motor state and cognitive demand. Thus, natural locomotion does not uniformly degrade hearing; instead, it preferentially down-weights a behaviorally critical, temporally demanding stage of auditory processing, offering a framework for understanding how motor-cognitive state shapes auditory scene analysis in the moving brain.

neuroscience↗

bioRxiv2025

Approaching an Error-Free Diploid Human Genome

Complete human genomes now resolve regions long inaccessible to genetic analysis, yet repetitive and structurally complex sequences remain prone to assembly errors because of limitations in current assembly algorithms. We introduce a support-based framework founded on the principle that each sequencing read provides independent evidence for the underlying genome sequence. Implemented through Sufficient Alignment Support (SAS), the framework identifies regions lacking concordant read support, localizes residual errors, and guides correction. Applying SAS to T2T-YAO, a haplotype-resolved Han Chinese genome, produced a support-validated assembly in which nearly all sequences outside unresolved rDNA arrays and long homopolymer tracts are supported by independent sequencing evidence. This work establishes a scalable framework for genome validation and provides a support-validated East Asian diploid reference for investigating human genomic diversity.

genomics↗

bioRxiv2025

Population-level super-pangenome reveals genome evolution and empowers precision breeding in watermelon

Pangenomes are increasingly critical for harnessing crop genetic diversity, yet their resolution and utility are often limited by insufficient sampling of high-quality genome assemblies. Here, we report a population-level watermelon super-pangenome constructed from 138 reference-grade assemblies, including 135 newly generated near-gapless genomes representing all seven watermelon species. The super-pangenome captures approximately one million structural variants (SVs), enabling accurate variant genotyping across ~900 watermelon accessions and substantially expanding variant discovery both across and within species. Broader sampling within the pangenome provides insights into genome evolution among watermelon species and sheds light on the origin of cultivated watermelon. SV-inclusive genome-wide association studies enhance trait mapping resolution and identify a copy number variation upstream of ClFCI1 that regulates flesh color intensity in a dosage-dependent manner. Leveraging this comprehensive variation map, we developed high-accuracy genomic prediction models for 18 agronomic traits. Together, our findings and genomic resources establish a foundational framework for dissecting complex traits and accelerating precision breeding in watermelon, while offering a valuable model for SV-resolved pangenomics in crop species.

plant biology↗

bioRxiv2025

ODFormer: a Virtual Organoid for Predicting Personalized Therapeutic Responses in Pancreatic Cancer

Pancreatic cancer (PC) patient-derived organoids (PDOs) faithfully recapitulate therapeutic responses but face clinical translation barriers, including high costs and technical complexity. To address these problems and the lack of frameworks for PDO-based drug-response assays, we developed ODFormer, a computational framework that simulates PC PDOs to predict clinically actionable, patient-specific drug responses by integrating transcriptomic and mutational profiles. ODFormer first employed two encoders, pretrained on 30,000 pan-cancer bulk transcriptomics and 1 million PC single-cell profiles respectively, to distil tissue-and organoid-specific representations. Then, trained on our curated 14,000 PDO drug-response assay (across 183 PDOs and 98 drugs) using a transformer-augmented hybrid contrastive network, ODFormer significantly outperformed state-of-the-art methods, notably achieving a PCC >0.9 in predicting standardized drug response. Multi-cohort retrospective analyses further demonstrated that ODFormer-guided personalized therapy significantly improves clinical outcomes, without requiring physical organoid assays. Furthermore, ODFormer identified novel clinico-biological PC subtypes and revealed therapy resistance biomarkers by stratifying predicted responders and non-responders. These were validated using independent datasets including TCGA-PDAC. Notably, ODFormer-guided treatment efficacy showed high concordance with prospective clinical responses by CA19-9.

bioinformatics↗

bioRxiv2025

Multimodal sensory overload in dopamine-deficient larval zebrafish leads to paradoxical kinesia

Paradoxical kinesia--the temporary alleviation of motor deficits by powerful, urgent stimuli in Parkinsons disease (PD)--remains poorly understood at the neural circuit level. Through chemo-genetic ablation of tyrosine hydroxylase-expressing neurons in larval zebrafish and brain-wide calcium imaging under head-fixed, tail-free conditions, we uncovered a neural mechanism underlying this phenomenon. While catecholamine (CA)-deficient larvae exhibited severe locomotor deficits during free swimming, they showed paradoxical recovery of tail movements during whole-brain neural activity imaging. This locomotor recovery was accompanied by a significantly increased number of active neurons in the midbrain and hindbrain, but with reduced firing rates. Further analyses across 2158 anatomically defined regions allowed us to uncover a subset of regions, genes, and neurotransmitter types. GABAergic neurons were found to primarily account for the hyperactivity in the hindbrain, while glutamatergic neurons accounted for the hyperactivity in the midbrain. Hierarchical clustering of neuronal activity with tail movements revealed distinct motor- and non-motor-associated hyperactive clusters in the hindbrain and midbrain, respectively. We identified the Mesencephalic Locomotor Region (MLR) sandwiched between these domains, with enhanced glutamatergic firing rate and cholinergic activation. Furthermore, we found that Telencephalic corticotropin-releasing factor b (crhb) expressing neurons play a crucial role in mediating stress-response to the tectum, which in turn triggers a cascade of neuronal hyperactivity downstream via MLR. These findings reveal a neural mechanism that links stress-induced sensory processing with motor control systems in the absence of regulatory feedback from catecholaminergic neurons, suggesting a direct, unmodulated pathway that bypasses typical inhibitory controls.

neuroscience↗

bioRxiv2025

APPLE-MS: A affinity purification-mass spectrometry method assisted by PafA-mediated proximity labeling

While affinity purification-mass spectrometry (AP-MS) has significantly advanced protein-protein interaction (PPI) studies, its limitations in detecting weak, transient, and membrane-associated interactions remain. To address these challenges, we introduced an innovative proteomic method termed Affinity Purification coupled Proximity LabEling-Mass Spectrometry (APPLE-MS), which combines the high specificity of Twin-Strep-tag enrichment with PafA-mediated proximity labeling. This method achieves unprecedented sensitivity while maintaining high specificity (4.07-fold over AP-MS). APPLE-MS also revealed the dynamic mitochondrial interactome of SARS-CoV-2 ORF9B during antiviral responses, while endogenous PIN1 profiling uncovered novel roles in DNA replication. Notably, APPLE-MS enabled in situ mapping of GLP-1 receptor complexes, demonstrating its unique capabilities for membrane PPI studies. This versatile method advances interactome research by providing comprehensive, physiologically relevant PPI networks, opening new opportunities for mechanistic discovery and therapeutic targeting. GRAPHICAL ABSTRACT O_FIG O_LINKSMALLFIG WIDTH=184 HEIGHT=200 SRC="FIGDIR/small/652320v1_ufig1.gif" ALT="Figure 1"> View larger version (53K): org.highwire.dtl.DTLVardef@aaf55borg.highwire.dtl.DTLVardef@1d25af1org.highwire.dtl.DTLVardef@150c1acorg.highwire.dtl.DTLVardef@db0d1c_HPS_FORMAT_FIGEXP M_FIG C_FIG MOTIVATION - Cell Reports Methods onlyAffinity purification-mass spectrometry (AP-MS) has become a widely used method for capturing high-affinity protein-protein interactions (PPIs). However, it still faces several limitations, including high levels of non-specific binding, challenges in detecting weak interactions, and limited ability to identify PPIs at the cell surface in situ. To overcome these challenges, we introduce a modular AP-MS method enhanced by PafA-mediated proximity labeling, which improves both the specificity and sensitivity of PPI detection in a single, streamlined workflow. SIGNIFICANCE - Cell Chemical Biology onlyProximity labeling is a powerful tool in chemical biology, providing critical support for studying protein function, cell signaling, and disease mechanisms by chemically labeling and capturing biomolecular interactions. In this study, we introduce a modular affinity purification-mass spectrometry (AP-MS) method that integrates proximity labeling. This method leverages the proximity-dependent enzymatic activity of PafA to covalently attach PupE to nearby proteins, in combination with the high-affinity interaction between Twin-Strep-tag and streptavidin to achieve efficient binding. This innovative approach enables precise identification of protein-protein interactions (PPIs), offering a powerful tool for capturing weak and transient interactions that are often missed by traditional AP-MS techniques. By enhancing both the specificity and efficiency of protein labeling, our work provides a robust chemical biology strategy to elucidate complex protein interactions, which are essential for understanding cellular processes and developing targeted therapeutic strategies, offering new insights into protein function and interaction dynamics in both health and disease.

biochemistry↗

bioRxiv2025

A synthetic biology toolkit for interrogating plasmid-dependent methylotrophy and enhancing methanol-based biosynthesis of Bacillus methanolicus

Bacillus methanolicus represents a thermophilic methylotroph whose methanol utilization depends on plasmid-encoded genes. It serves as a unique model for deciphering plasmid-dependent methylotrophy and an ideal chassis for low-carbon biomanufacturing using CO2-derived C1 substrates. Despite its evolutionary uniqueness and industrial potential, the lack of synthetic biology tools has hindered both mechanistic understanding and strain engineering. Here, we present a comprehensive synthetic biology platform comprising a high-efficiency electroporation protocol, a CRISPR method enabling robust and multiplex genome editing, diverse neutral loci for gene integration and overexpression, and a cloud-based genome-scale metabolic model iBM822 for user-friendly biodesign. Leveraging this toolkit, we systematically dissected plasmid-dependent methylotrophy, host restriction-modification systems, and functional significance of the chromosomal methylotrophic genes through targeted deletion. To address plasmid loss-induced strain degeneration, we integrated the large endogenous plasmid pBM19 into the chromosome for stable and intact methylotrophic growth. Finally, by integrating metabolic modeling with CRISPR editing, we engineered L-arginine feedback regulation to achieve the first L-arginine biosynthesis from methanol. This study establishes a synthetic biology framework for B. methanolicus, promoting mechanistic exploration of methylotrophy and low-carbon biomanufacturing.

synthetic biology↗

bioRxiv2025

Control of leaflet movement pattern by a novel PP2C phosphatase DLM1 in Medicago truncatula

Sessile plants exhibit diverse movement behaviors that have long intrigued the scientists. The legume plants display a rhythmic leaflet movement pattern characterized by horizontal opening during the day and vertical closure at night. However, the underlying mechanisms remain largely enigmatic. Here, we isolated a mutant designated as dlm1 (downward leaflet movement1) from Medicago truncatula that displays leaflets downward opening during daytime while upward closure at night. Cellular analyses reveal that this aberrant phenotype correlates with abnormal volume changes of motor cells within the pulvinus. DLM1 exhibits high expression level in the motor organ pulvinus and encodes an unreported nuclear-localized PP2C phosphatase. Notably, phylogenetic analysis demonstrates that species exhibiting rhythmic leaflet movements consistently retain DLM1 homologs, suggesting the possibility of its functionally conserved role in regulating leaflets movement pattern. Structural characterization reveals that DLM1 possesses both phosphatase and kinase domains. Functional complementation assays demonstrate that the phosphatase domain is necessary and sufficient for maintaining the leaflet movement pattern. Collectively, our work uncovers a novel PP2C protein that governs the leaflet movement, providing mechanistic insights into this intriguing plant behavior.

plant biology↗

bioRxiv2025

Spatiotemporal Analysis of Remyelination Reveals a Concerted Interferon-Responsive Glial State That Coordinates Immune Infiltration

Remyelination, the process by which axons are re-encased in myelin after injury, is a critical step in restoring brain function, yet the dynamics from initial injury to repair remain poorly characterized. Here, we combined optimized single-nucleus RNA-seq with Slide-seqv2, a high-resolution spatial transcriptomics technology, to densely reconstruct the cellular processes that coordinate remyelination after a focal demyelinating injury. This revealed several findings: First, we found extensive transcriptional diversity of glia and monocyte-derived macrophages from demyelination to repair. Second, we identified a population of infiltrating peripheral lymphocytes--predominantly CD8 T-cells and natural killer cells--that are enriched specifically during remyelination. Third, we identified a concerted interferon-response gene signature that is shared across several cell types--microglia, astrocytes, and the oligodendrocyte lineage--just prior to reestablishment of myelin. These interferon-responsive glia (IRG) form clusters around remyelinating white matter and their formation is solely dependent on the type I interferon receptor. Functionally, we found that IRG secrete the cytokine CXCL10 which mediates infiltration of peripheral lymphocytes into the repairing white matter. Depletion of the most abundant infiltrating lymphocyte, CD8 T-cells, attenuated the differentiation of mature oligodendrocytes during remyelination. Together, our data reveals the diversity of glial-immune interactions that orchestrate white matter repair and a type I-dependent glial state that drives lymphocyte influx into damaged white matter to modulate oligodendrocyte differentiation. O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=57 SRC="FIGDIR/small/649486v1_ufig1.gif" ALT="Figure 1"> View larger version (27K): org.highwire.dtl.DTLVardef@2c7e07org.highwire.dtl.DTLVardef@1063719org.highwire.dtl.DTLVardef@10b8862org.highwire.dtl.DTLVardef@26dd8b_HPS_FORMAT_FIGEXP M_FIG C_FIG

neuroscience↗

bioRxiv2025

ReCellTy: Domain-specific knowledge graph retrieval-augmented LLMs workflow for single-cell annotation

To enable precise and fully automated cell type annotation with large language models (LLMs), we developed a graph-structured feature-marker database to retrieve entities linked to differential genes for cell reconstruction. We further designed a multi-task workflow to optimize the annotation process. Compared to general-purpose LLMs, our method improves human evaluation scores by up to 0.21 and semantic similarity by 6.1% across 11 tissue types, while more closely aligning with the cognitive logic of manual annotation.

bioinformatics↗

bioRxiv2025

Spatially aligned random partition models on spatially resolved transcriptomics data

We propose spatially aligned random partition (SARP) models for clustering multiple types of experimental units, incorporating dependence in a subvector of the cluster-specific parameters, e.g., a subvector of spatial information, as in the motivating application. The approach is developed for inference about co-localization of immune, stromal, and tumor cell sub-populations. The aim is to understand the recruitment of immune and stromal cell subtypes by tumor cells, formalized as spatial dependence of the corresponding homogeneous cell subpopulations. This is achieved by constructing Bayesian nonparametric random partition models for the different types of cells, with a hierarchically structured prior introducing the desired dependence. Specifically, we use Pitman-Yor priors and add dependence in the base measure for spatial features, while leaving the base measure corresponding to gene expression features a priori independent across different types of cells. Details of the model construction are designed to lead to a convenient MCMC algorithm for posterior inference. Simulation studies show favorable performance in identifying co-localization between types of cells. We apply the proposed approach with colorectal cancer (CRC) data and discover subtypes of immune and stromal cells that are spatially aligned with specific tumor regions.

bioinformatics↗

bioRxiv2025

Targeting protein glycosylation and cholesterol metabolism in chemoresistant pancreatic cancer

Chemotherapy remains the cornerstone of pancreatic ductal adenocarcinoma (PDAC) treatment. However, most cases of PDAC finally progress to advanced chemoresistant disease. This highlights an urgent need to develop new pharmacological strategies to overcome chemotherapy resistance using clinical grade inhibitors. Here, we established a biobank comprising 260 organoid lines derived from 322 pancreatic cancer patients. These organoids underwent extensive multi-omics profiling and drug sensitivity testing for both chemotherapy and targeted therapy. Increased protein glycosylation and cholesterol metabolism signaling pathways were especially enriched in chemoresistant organoid lines. Importantly, chemoresistant PDAC organoids were effectively targeted with statins, inhibitors of cholesterol synthesis. Statins treatment attenuated protein glycosylation, cholesterol levels, and the epithelial-to-mesenchymal transition (EMT) signature in PDAC organoids. We further conducted a phase 2 clinical trial (NCT06241352) combining atorvastatin with chemotherapy in patients with advanced chemoresistant pancreatic cancer. Among 29 patients, 21 (72.4%) demonstrated a response, with tumor markers decreasing by more than 20%, suggesting durable responses and potential clinical benefits in this challenging patient population. ClinicalTrials.gov identifier: NCT06241352.

cancer biology↗

bioRxiv2025

Genetic association between posterior parietal cortex and intelligence

The identification of intelligence-associated loci raises questions for functional association of brain regions and for different intelligence concepts of either a general capability or discrete abilities. By generating single-nucleus transcriptomic atlas of mouse posterior parietal cortex (PPC), which is implicated in quantification behavior, we find that human intelligence-associated gene homologs are significantly enriched in certain specific neurons within the PPC. We also find 73% of the human intelligence gene homologs are highly enriched in PPC. In addition to PPC, several regions at different locations, not cortex next to PPC, are also significantly associated with intelligence genes. The separate locations of these cortex regions and their regional-specific expressed intelligence genes raise the possibility for discrete intelligence capabilities. Animal behavior analysis reveals that mutations for two top intelligence-associated genes with specific expression in either PPC or entorhinal cortex (ENT) affect quantification or learning and memory, respectively. Finally, five patients with PPC resection in the left hemisphere also display quantification defects, but not other cognitive abilities. Together, our studies provide the first genetic and functional evidence for the involvement of PPC in intelligence. Our data also support the model of multiple intelligence over a general capability for intelligence.

genetics↗

bioRxiv2025

Benchmarking AlphaFold3-like Methods for Protein-Peptide Complex Prediction

Protein-short peptide interactions are central mechanisms regulating signal transduction, chaperone functions, and drug targeting design. However, the precise prediction of their complex structures has long been constrained by the limited accuracy of traditional methods. The breakthrough of AlphaFold2 resolved the challenge of predicting protein monomer structures. Its derivative method, AlphaFold2-multimer, surpassed traditional docking approaches in predicting protein complexes, yet its accuracy remains suboptimal. AlphaFold3 and its replication models, such as Protenix, Chai-1, and Boltz-1, have further improved prediction accuracy for protein-protein interactions. To evaluate the performance of these methods in protein-short peptide structure prediction, this study conducted predictions and comprehensive comparisons on two benchmark datasets of protein-short peptide complexes. One dataset did not intentionally exclude structures from the training set of these methods, while the other did. For the former, the results demonstrate that the new-generation structural method significantly increased the success rate under stringent criteria (Fnat[≥] 0.8) to 70-80%, compared to AlphaFold2-multimer (53%), with Protenix achieving the highest accuracy at 80.8%. For the second dataset, AlphaFold2-multimer showed consistent performance, but the success rates of the new-generation methods all dropped significantly (to 40-56%). Our analysis suggests that the main challenge for the performance decline comes from the accurate prediction of binding sites. For all datasets, A multi-method combination strategy enabled a higher-quality prediction success rate compared to the individual models, highlighting synergistic advantages among methods. Notably, inter-model consistency analysis provided an efficient metric for selecting optimal predictions.

bioinformatics↗

bioRxiv2025

Small Molecule Modulators of TOX protein to Re-invigorate T cell Activity

The TOX protein (thymocyte selection-associated high mobility group box) is a critical transcription factor implicated in both T acute lymphoblastic leukemia (T-ALL) and CD8+ T cell exhaustion. Gene perturbation studies suggest that inhibiting TOX may have therapeutic implications for both leukemia and T cell exhaustion. However, due to its complex molecular mechanisms and intrinsically disordered structure, TOX has not been effectively targeted by small molecules to date. In this study, we used small molecule microarray (SMM) screening and biochemical assays to identify a series of TOX protein-protein interaction (PPI) inhibitors. We identified KI-TOX-A3 as a TOX protein binder and potent TOX PPI inhibitor. In T-ALL, KI-TOX-A3 revealed selective cytotoxicity and proteosome-dependent TOX degradation. In CD8+ T cells, KI-TOX-A3 potently reversed T cell exhaustion by decreasing surface inhibitory receptors, increasing expression of effector cytokines, and enhancing cancer cell killing activity. We also demonstrate the utility of KI-TOX-A3 to probe potential epigenetic regulatory mechanisms of TOX via KAT7 acetylation in T cells.

biochemistry↗

bioRxiv2025

FACED 2.0 enables large-scale voltage and calcium imaging in vivo

Monitoring neuronal activity at large scale and high spatiotemporal resolution is crucial for understanding information processing within the brain. We optimized a kilohertz-frame-rate two-photon fluorescence microscope with all-optical megahertz line-scan rate to achieve ultrafast imaging across large areas and volumes at subcellular resolution. Applying this technique to voltage and calcium imaging in vivo, we demonstrated simultaneous recording of voltage activity over 200 neurons and calcium activity over 14,000 neurons.

neuroscience↗

bioRxiv2025

Development of an image-guided non-vitrectomy subretinal access approach for trans-scleral cell and gene therapy delivery

PurposeRegenerative therapies for retinal diseases include cell and gene therapy modalities that are targeted to the subretinal space. Several recent clinical trials have shown that the morbidity of surgical access is the major limitation of safe subretinal space delivery. We aimed to develop an image-guided procedure for minimally invasive subretinal access (MISA) as a platform to deliver therapeutic agents for the treatment of degenerative retinal diseases. MethodsWe engineered prototypes of a novel common-path swept source optical coherence tomography (CP-SSOCT)-enabled needle, coaxial guide (COG), and subretinal access cannula (SAC). We pilot tested the MISA procedure in ex vivo bovine eyes and in vivo porcine ocular surgery. ResultsA- and M-mode scan recordings of ex vivo and in vivo animal eye models demonstrated that CP-SSOCT imaging from the scleral side (ab externo) was capable of identifying the retinal laminae and the sub-retinal space. We show results from in vivo porcine MISA surgeries (N=4) using the novel CP-SSOCT-enabled sub-retinal injection needle, COG, and SAC through the transscleral approach. The MISA approach enabled subretinal device placement in the posterior pole, however, cases of retinal incarceration and retinal perforation were encountered. ConclusionsWe describe a novel CP-SSOCT-guided subretinal access approach that, with further optimization, may be useful in regenerative retinal surgery.

bioengineering↗