Biology subjects

Zou, J.

Publications and source records attributed to Zou, J..

At least 37 records · Page 2Linked to original sources

bioRxiv2026

Single-cell transcriptomic atlas of mouse oocyte development from growth to ovulation

Oocyte growth and maturation depend on tightly coordinated programs within oocytes and their surrounding granulosa cells, yet defining the transcriptional continuum of growing oocytes has been challenging due to their large size and the limitations of current droplet-based and spatial transcriptomic platforms. Here, we optimized a high-density array-based platform called Stereo-cell for high-throughput dual-modality profiling of large mouse oocytes, allowing for both the preservation of morphology and transcript capture. By integrating unsupervised transcriptomic clustering with cell morphological features, we delineated successive temporal windows from growing oocytes to metaphase II and uncovered stage-linked shifts from early programs toward later programs. We also profiled the ovarian somatic fraction, reconstructed granulosa-cell subtype relationships, and placed ovarian cell states in tissue context using single-cell-resolution spatial data.

developmental biology↗

bioRxiv2026

Eubiota: Modular Agentic AI for Autonomous Discovery in the Gut Microbiome

The gut microbiome regulates many aspects of human biology, including immunity and inflammatory diseases, yet mechanistic discovery and translation remain constrained by fragmented workflows and manual hypothesis integration. Here, we present Eubiota, an open-source modular agentic framework that decomposes microbiome inquiry using specialized agents for planning, execution, verification, and grounded synthesis. Coordinated through shared memory and domain-specific tools, Eubiota employs reinforcement learning to optimize multi-turn reasoning, achieving 87.7% benchmark accuracy and outperforming GPT-5.1 by 10.4%. Across four case studies, Eubiota enabled end-to-end discovery with experimental validation. Screening nearly 2,000 bacterial genes, Eubiota identified the uvr-ruv DNA repair axis as a fitness determinant under inflammatory stress, validated using transposon mutants and IBD metagenomes. It further designed a four-strain consortium that attenuated colitis severity in mice and generated a commensal-sparing antibiotic cocktail, demonstrating its utility in addressing community-level design challenges at cellular and molecular levels. Finally, Eubiota discovered diet-associated metabolites that suppress NF-{kappa}B signaling. Together, these results establish Eubiota as a scalable, tool-grounded scientific copilot for mechanistically driven microbiome discovery.

bioinformatics↗

bioRxiv2026

Vagus Nerve Stimulation Ameliorates Epilepsy via a Gut Microbiota-Butyrate-Vagus Nerve-Brain Axis

Vagus nerve stimulation (VNS) is an established therapy for drug-resistant epilepsy, yet its mechanism remains partially understood. Here, through an integrated clinical-translational approach, we demonstrate that VNS modulates the gut microbiome and alter the level of butyrate to reduce seizure susceptibility via vagus nerve-dependent signaling. In a longitudinal cohort of epilepsy patients, VNS enriched short-chain fatty acid-producing bacteria, including Akkermansia muciniphila, Prevotella, Alistipes, and Odoribacter. Among clinical responders, fecal butyrate levels decreased concomitantly with an increase in plasma butyrate concentrations. Fecal microbiota transplantation from post-VNS patients into mice reduced seizure severity and improved gut barrier integrity. In a murine model of epilepsy, VNS similarly increased the abundance of Akkermansia and plasma butyrate concentrations, decreased fecal butyrate levels, and abolished its antiseizure effects upon vagotomy. Either colonization with Akkermansia muciniphila or exogenous butyrate supplementation was sufficient to reduce seizures and anxiety-like behavior, effects that also required intact vagal signaling. Our findings define a microbiota-butyrate-vagal pathway essential for the therapeutic effects of VNS, highlighting the gut-brain axis as a target for epilepsy treatment. GRAPHICAL ABSTRACT O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=200 SRC="FIGDIR/small/707138v1_ufig1.gif" ALT="Figure 1"> View larger version (75K): org.highwire.dtl.DTLVardef@8243e6org.highwire.dtl.DTLVardef@37f0e6org.highwire.dtl.DTLVardef@a8b321org.highwire.dtl.DTLVardef@1887342_HPS_FORMAT_FIGEXP M_FIG C_FIG HIGHLIGHTSO_LIVagus nerve stimulation (VNS) remodels the gut microbiota and alter the level of butyrate. C_LIO_LIVNS enriches short-chain fatty acids-producing bacteria, specifically Akkermansia muciniphila, in both humans and mice. C_LIO_LITransplantation of VNS-conditioned microbiota confers seizure protection and restores gut barrier integrity. C_LIO_LIAkkermansia muciniphila or exogenous butyrate is sufficient to reduce seizures via a vagus nerve-dependent pathway. C_LI

neuroscience↗

bioRxiv2026

The Virtual Biotech: A Multi-Agent AI Framework for Therapeutic Discovery and Development

Drug discovery and development requires integrating diverse evidence across biological scales and data modalities. However, relevant data, tools, and expertise remain fragmented across teams and organizations, making integration difficult. To address these challenges, we introduce the Virtual Biotech, a coordinated team of AI agents that mirrors the structure of human therapeutic research organizations to support end-to-end computational discovery. The Virtual Biotech is led by a Chief Scientific Officer agent that receives scientific queries, delegates them to domain-specialized scientist agents, and integrates their outputs through data-driven reasoning. Scientist agents leverage complementary tools and knowledge sources spanning statistical genetics, functional genomics, pathways and interactions, chemoinformatics, disease biology, and clinical data. We showcase the Virtual Biotech across three translational applications. First, the agents autonomously annotated and analyzed outcomes from 55,984 clinical trials to identify genomic features of drug targets associated with trial success. More than 37,000 clinical-trialist agents curated structured trial outcomes and linked targets to multi-omic annotations, including cell-type-specific features derived by the agents from single-cell RNA-sequencing atlases. The agents discovered that drugs targeting cell-type-specific genes were 40% more likely to progress from Phase I to Phase II and 48% more likely to reach market (Phase IV), while exhibiting 32% lower adverse event rates. Second, the Virtual Biotech evaluated B7-H3 as a lung cancer target, integrating statistical genetics, single-cell, spatial, and clinicogenomic evidence to propose an antibody-drug conjugate strategy while identifying key liabilities and differentiation opportunities. Third, the platform analyzed a terminated ulcerative colitis trial targeting OSMR{beta} to infer potential failure mechanisms and proposed biomarker-guided enrollment strategies to address precision-medicine gaps. Together, these results illustrate how the Virtual Biotech can enable more transparent, efficient, and comprehensive multi-scale therapeutic analyses, helping to accelerate early-stage drug discovery workflows while keeping human scientists in the loop.

genomics↗

bioRxiv2026

AI-driven discovery and engineering of human endogenous nanocage proteins for mRNA delivery

Safe and effective gene delivery remains a central challenge for therapeutic applications. While non-viral and viral vectors have enabled substantial progress, their reliance on non-human components often triggers immune responses, limiting their use in chronic treatments. Here, we developed DeepDelivery, an artificial intelligence-driven platform to repurpose human proteins for mRNA delivery. An unbiased screening of the human proteome nominated 512 candidates, with experimental validation confirming that 80% of top-ranked hits form mRNA-encapsulating particles and mediate efficient functional delivery in human cells without provoking detectable inflammation. Notably, multiple tripartite motif (TRIM) family proteins, typically linked to antiviral responses, exhibited strong assembly and delivery activity. Quantitative analysis and interpretation of the model revealed structural domains that govern nanocage formation, enabling domain-guided engineering of TRIM25 variants with enhanced function. Our work establishes a generalizable framework for discovering human-derived delivery vehicles and provides a path toward programmable, non-immunogenic mRNA therapeutics.

synthetic biology↗

bioRxiv2026

Functional and Structural Basis of Omicron BA.3.2.1 Spike

SARS-CoV-2 Omicron descendant BA.3.2, derived from BA.3 and carrying 39 additional spike substitutions, prompts concerns about altered fitness and antigenicity. Using BA.3.2.1 as a representative sublineage, we have engineered live-attenuated mNeonGreen SARS-CoV-2 encoding spikes from BA.3.2.1 and the JN.1-derived LP.8.1 and XEC, and benchmarked them against BA.3 and KP.3 in Calu-3 cells and primary human airway epithelium. BA.3.2.1 outcompetes BA.3 yet replicates more slowly than JN.1-descendants. Despite increased RBD-hACE2 affinity, BA.3.2.1 spike binds hACE2 less efficiently than LP.8.1, and shows the greatest resistance to neutralization by sera from KP.2/KP.3 infections. Cryo-EM reveals a predominantly closed, compact, asymmetric trimeric spike with a remodeled N-terminal domain, a distinct fusion-peptide-proximal region and altered N-glycosylation. BA.3.2.1 spike adopts a fine-tuned receptor-binding interface and shifted monoclonal-antibody footprints. These findings identify a trade-off between immune evasion and replication fitness driven by closed spike conformation, rationalizing BA.3.2s limited prevalence and underscoring the necessity for continued variant surveillance.

microbiology↗

bioRxiv2026

The amplitude and latency of the earliest signal in V1 encode bottom-up saliency by feature conjunction

The neural origin of bottom-up saliency for exogenous attention remains highly controversial. In this study, we investigated whether the earliest activity in the primary visual cortex (V1) encodes saliency signals defined by the eye-of-origin and feature-conjunction information. Electroencephalography (EEG) recordings from the human occipital cortex revealed early responses to eye-of-origin (E) and/or orientation (O) singletons, with larger response amplitudes to the double-feature (EO) singletons. The short onset latency (58-70 ms) and polarity reversal of the responses indicate a V1 origin. Importantly, the latency and amplitude of these responses predicted behavioral detection performance. Together, these findings suggest that the timing and amplitude of the earliest signals in V1 represent the saliency of combined feature contrasts for bottom-up attention. These signals unlikely originate from projections of other proposed source areas of saliency, due to the scarcity of necessary monocular neurons to process eye-of-origin information. HighlightsO_LIEye-of-origin information, invisible to the SC, elicits an early saliency signal in V1 within 50-100 ms. C_LIO_LICombined feature contrast enhances V1 saliency responses in a nonlinear fashion. C_LIO_LIThe latency and amplitude of V1 saliency responses predict behavioral detection performance. C_LI

neuroscience↗

bioRxiv2026

Distinct subcortical pathways mediate collision detection with and without attention and consciousness

Accurate detection of impending collisions is essential for survival. While multiple subcortical pathways have been found in the detection of looming stimuli, the roles of top-down attention and consciousness in this process remains unclear. Using high-resolution fMRI, we investigated how top-down attention and conscious awareness influence the subcortical processing of collision trajectories in healthy participants and hemianopic patients with cortical blindness. In healthy participants, the superior colliculus (SC)-parabigeminal nucleus (PBGN)-amygdala pathway integrated binocular information for accurate detection of collision trajectories in the upper visual field, but only when participants were paying attention to the looming stimuli. On the other hand, for stimuli presented in the blind visual field of hemianopic patients, effective connectivity analyses revealed collision sensitivity in the SC-pulvinar and SC- ventral tegmental area (VTA) pathways, supporting automatic collision detection even in the absence of visual awareness. Together, these findings suggest that distinct subcortical pathways mediate collision detection in the human brain depending on the presence and absence of attention and conscious awareness. HighlightsO_LIAttention selectively enhanced collision sensitivity in the SCs-PBGN-CMA pathway. C_LIO_LIBinocular information contributed to accurate collision detection in the upper visual field. C_LIO_LISC-Pulvinar and SC-VTA pathways detect collision trajectory without visual awareness. C_LI Significance statementAccurate detection of collision trajectory is critical for survival. Using high-resolution fMRI in healthy participants and hemianopic patients, we show that collision detection is mediated by distinct subcortical circuits depending on the attentional state and conscious awareness. The SC-PBGN-amygdala pathway can precisely detect collision trajectory in upper visual field with top-down attention, whereas the SC-pulvinar and SC-VTA pathways enable automatic collision detection without awareness. These findings reveal distinct functional roles of the subcortical pathways in collision detection.

neuroscience↗

bioRxiv2025

Template-based RNA structure prediction advanced through a blind code competition

Automatically predicting RNA 3D structure from sequence remains an unsolved challenge in biology and biotechnology. Here, we describe a Kaggle code competition engaging over 1700 teams and 43 previously unreleased structures to tackle this challenge. The top three submitted algorithms achieved scores within statistical error of the winners of the recent CASP16 competition. Unexpectedly, the top Kaggle strategy involved a pipeline for discovering 3D templates, without the use of deep learning. We integrated this template-modeling pipeline and other Kaggle strategies to develop a single model RNAPro that retrospectively outperformed individual Kaggle models on the same test set. These results suggest a growing importance of template-based modeling in RNA structure prediction.

biophysics↗

bioRxiv2025

Nuclear export governs TDP-43 phase transitions and cytoplasmic aggregation

RNA-binding protein TAR DNA-binding protein 43 (TDP-43) can form liquid-like, nuclear assemblies whose phase behavior is thought to influence its aggregation propensity and neurotoxic activity. The mechanism(s) that governs the liquid-to-solid phase transition of TDP-43 is poorly defined. Here we combined chemical and genome-wide genetic screenings to identify cellular factors that modulate the phase behavior of an RNA-binding defective TDP-43 mutant. Our screens uncovered multiple cellular processes including RNA splicing, protein translation, proteostasis imbalance and nuclear export as TDP-43 phase regulators. We also developed a semi-permeabilized cell system that partially recapitulates the TDP-43 phase transition in vitro, and showed that nuclear export inhibition reshapes the nuclear environment to favor RNA-dependent liquid-liquid phase separation (LLPS) of TDP-43, which mitigates its aggregation. Nuclear export inhibition in a brain organoid model bearing an ALS-associated mutation reduces pathogenic phospho-TDP-43 accumulation. These findings identify multiple modulators of TDP-43 phase transitions in a sensitized model system and establish a framework for further dissecting the link between nuclear transport and TDP-43 phase dynamics.

cell biology↗

bioRxiv2025

Modeling patient tissues at molecular resolution with Eva

Tissue structure is essential to function and homeostasis in all organs, and disruptions to structure usually indicate disease. Modeling relationships between structural, molecular, and clinical aspects of tissues could advance new diagnostics and treatment strategies. Although profiling techniques like spatial proteomics can capture these relationships, the data remain challenging to extract insight from. Here, we present Eva, a foundation model for tissue imaging data that learns multi-scale spatial representations of tissues at the molecular, cellular, and sample level. Eva uses a novel vision transformer architecture and is pre-trained on masked reconstruction of over 40 million matched spatial proteomics and histopathology images. We show that Eva excels at a variety of tasks, including cross-modal inference from H&E to proteomics stains, quality control, data annotation, zero-shot retrieval, survival modeling, and patient stratification. Extensive evaluations on held-out validation data demonstrate the versatility and generalizability of the learned embeddings. We anticipate that Eva will accelerate translational science by bridging basic research and clinical practice.

bioinformatics↗

bioRxiv2025

Neddylation Regulates Mitochondrial Dynamics and Turnover in the Adult Heart

BACKGROUNDDisruption of mitochondrial homeostasis drives cardiomyopathy and heart failure, yet upstream regulatory mechanisms remain poorly defined. Neddylation, a reversible post-translational conjugation of the ubiquitin-like protein NEDD8 by E1/E2/E3 enzymes, is essential for cardiac morphogenesis, but its role in the adult heart is unknown. METHODSWe assessed the relevance of neddylation to human cardiac disease by gene set enrichment analysis of ischemic (ICM) and non-ischemic cardiomyopathy (NICM) datasets and by immunoblotting and qPCR of ventricular tissue from patients with ICM or dilated cardiomyopathy (DCM). In adult mice, we induced cardiomyocyte-restricted deletion of the NEDD8-activating enzyme 1 (NAE1) by tamoxifen injection and monitored cardiac function at baseline and after transverse aortic constriction (TAC). Bulk RNA-seq 4 weeks post-tamoxifen was combined with bioenergetic, biochemical, and ultrastructural analyses. To assess mitochondrial dynamics, we generated NAE1/MFN2 and NAE1/DRP1 double-knockout mice. Cullin activity, mitochondrial ubiquitination, and mitophagy were measured in hearts and cultured cardiomyocytes. RESULTSNeddylation pathways were dysregulated in human ICM and NICM datasets and in failing ICM/DCM myocardium. Cardiomyocyte-specific NAE1 deletion caused systolic dysfunction and heart failure by 10 weeks post-tamoxifen, culminating in premature death and exacerbating TAC-induced pressure-overload heart failure. At 4 weeks, NAE1 loss repressed metabolic and mitochondrial bioenergetic programs, reduced ATP production, and impaired respiration. Electron microscopy revealed elongated mitochondria and accumulated mitophagic vesicles, with dysregulation of DRP1, MFN2, PINK1, LC3-II, and p62. DRP1/NAE1 co-deletion accelerated systolic failure relative to either single knockout, whereas MFN2/NAE1 co-deletion did not alter early disease progression, implicating pathogenic mitochondrial hyperfusion. Genetic NAE1 depletion in vivo and pharmacologic NAE1 inhibition in vitro impaired mitophagic vesicle formation and flux, inactivated cullin scaffold proteins, reduced mitochondrial ubiquitination, and blunted mitophagic clearance. CONCLUSIONSCardiac neddylation preserves adult heart function by coordinating mitochondrial fusion-fission dynamics and sustaining cullin-dependent ubiquitination and turnover of damaged mitochondria. These findings identify neddylation as a key regulator of mitochondrial quality control and link its disruption to human cardiomyopathy. Therapeutically, targeting the neddylation-cullin axis may limit mitochondrial dysfunction, enhance mitophagy, and improve energetic reserve in failing hearts, while neddylation signatures in patient myocardium may help guide stratification and precision therapy for cardiomyopathy. Clinical PerspectiveO_ST_ABSWhat Is New?C_ST_ABS* Demonstrates for the first time that the NEDD8-activating enzyme (NAE1)driven neddylation pathway is indispensable for maintaining mitochondrial quality control in the adult heart. * Links loss of neddylation to mitochondrial hyperfusion, impaired mitophagy, and rapid progression to heart failure. * Reveals that neddylation promotes cullin-RING ligase-mediated ubiquitination of damaged mitochondria, coupling mitochondrial dynamics with turnover. What Are the Clinical Implications?* Restoring or enhancing cardiac neddylation may represent a novel therapeutic avenue for cardiomyopathies characterized by mitochondrial dysfunction. * Pharmacologic agents that bolster DRP1-dependent fission or activate cullin neddylation could potentially normalize mitochondrial dynamics and improve myocardial energetics. * Conversely, systemic neddylation inhibitors now in oncology trials warrant careful cardiac monitoring, as they may precipitate mitochondrial injury and heart failure. * Circulating or tissue markers of neddylation might help stratify patients at heightened risk for mitochondrial-driven cardiac disease and guide precision therapy.

physiology↗

bioRxiv2025

BrainBridge Characterizes Key Factors affecting Alzheimer's Disease and Associated Phenotypes

Single-cell RNA sequencing (scRNA-seq) has significantly advanced our understanding of Alzheimers disease and aging by revealing cellular heterogeneity and shifts in cell-type composition between diseased/old and healthy/young individuals. However, few existing studies utilize the rich information in single-cell transcriptomic atlases for robust patient-level modeling and biological feature selection. To address this gap, we present BrainBridge, a deep learning-based framework designed to integrate atlas-scale single-cell transcriptomic data with phenotypic information to model the biomolecular complexity of the human brain. BrainBridge functions both as a powerful predictor and an embedding model for representing sample-level expression profiles and covariates through comprehensive benchmarking. We also demonstrate its effectiveness in prioritizing key genes and cell types associated with disease progression, aging, and sex differences. We further validate our findings using external resources, including genome-wide and epigenome-wide association studies (GWAS and EWAS), spatial transcriptomics, and perturb-seq experiments. Finally, we deploy BrainBridge within an interactive, agent-powered interface that enables intuitive and user-friendly model interactions, promoting broader accessibility and application in biomedical research.

bioinformatics↗

bioRxiv2025

SpatialProp: tissue perturbation modeling with spatially resolved single-cell transcriptomics

Perturbational studies are the gold standard for identifying causal relationships between components of biological systems. Recent technological advances, including Perturb-seq and related assays, have enabled high-throughput screening of genetic perturbation effects on single cells. Several machine learning tools have also been developed to infer the effect of single-cell perturbations. However, both approaches are generally limited to dissociated cells, and the effect of genetic perturbations on neighboring cells within intact tissue has not yet been explored. Here we introduce a computational framework using graph neural networks for predicting the effect of multi-gene, multi-cell type perturbations on cells in whole tissue sections. We leverage the natural heterogeneity in tissue microenvironments across spatially resolved single-cell transcriptomics datasets to train SpatialProp (Spatial Propagation of Single-cell Perturbations). We show that SpatialProp can predict gene expression from the tissue microenvironment and map fine-grained steering of tissue microenvironments to new target states. To assess for causal enrichment in spatial perturbation predictions, we propose CausalInteractionBench, a bidirectional benchmarking approach using curated cell-cell interactions. Under this benchmark, we evaluate the causal utility of SpatialProp in predicting the spatial effects of different perturbations. SpatialProp provides a framework towards rapid hypothesis generation and in silico perturbation experiments, particularly in the study of spatially patterned tissue biology.

bioinformatics↗

bioRxiv2025

A human forebrain organoid model phenocopies dysregulated RNA and protein homeostasis in ALS/FTD-associated TDP-43 proteinopathies

TAR DNA-binding protein 43 (TDP-43) proteinopathy is a defining pathological feature of amyotrophic lateral sclerosis (ALS) and frontotemporal dementia (FTD), yet the downstream molecular events linking TDP-43 dysfunction to neurodegeneration remain incompletely understood. Here, we model TDP-43 proteinopathy by introducing the ALS/FTD-associated TARDBP K181E mutation into human induced pluripotent stem cells and differentiating them into three-dimensional forebrain organoids. Organoids carrying the K181E mutation exhibit spontaneous TDP-43 hyperphosphorylation, cytoplasmic p-TDP43 accumulation, RNA dysregulation, pro-inflammatory signaling, and activation of apoptotic pathways without TDP-43 overexpression or external stress. Single-cell transcriptomics and enhanced crosslinking immunoprecipitation reveal that the K181E mutation alters TDP-43 RNA-binding specificity, causing widespread RNA mis-splicing including cryptic exon inclusion. Among affected targets, we identify cryptic exon inclusion in the transcription factor PRDM2. PRDM2 cryptic exon-derived peptide immunoreactivity was detected in ALS spinal motor neurons and associated with p-TDP-43 pathology. These findings identify PRDM2 mis-splicing as a candidate downstream event linking TDP-43 dysfunction relevant to ALS and FTD. One sentence summaryHuman brain organoids reveal how mutant TDP-43 disrupts RNA processing and causes neuronal stress.

neuroscience↗

bioRxiv2025

Elevated Temperatures Representing Heatwave Conditions Shift Active Nitrifying Communities and Their Viruses in Tidal Flats and Agricultural Soils

Global heatwave intensification associated with climate change will impact the nitrogen cycle, yet its effect on specific nitrifier groups or their interactions with viruses remains unclear. Using 13CO2-DNA-based stable isotope probing (SIP) coupled with metagenomics, we show that elevated temperatures associated with heatwave conditions restructure active nitrifying communities and their viruses in Yangtze River estuary tidal flats and adjacent agricultural soils. In tidal flats, high temperatures shifted active ammonia-oxidizing archaea and bacteria (AOA and AOB), and nitrite-oxidizing bacteria (NOB) from marine to terrestrial ecotypes. In contrast, heatwave conditions stimulated terrestrial ecotypes of AOA but suppressed AOB in agricultural soils. 13C-labeled nitrifier-infecting viruses also showed temperature-driven shifts in activity, lifestyle, and auxiliary metabolic genes in concert with hosts. Notably, AOA viruses carried the plastocyanin gene, potentially augmenting host metabolism. These findings demonstrate heatwaves drive complex shifts in nitrifier communities and their interactions with viruses, impacting global nutrient cycling under climate extremes.

microbiology↗

bioRxiv2025

Enhancing CRISPR Education: A Plug-And-Play Framework for Teaching Gene Editing in Plant Biology

Modern molecular biology tools and technologies such as CRISPR have sped up scientific discovery. From an educational perspective, these advancements are both exciting and overwhelming. Educators shaping these future scientists face the ongoing challenge of staying compliant with the latest developments in molecular biology while finding effective ways to teach these discoveries. As the use of CRISPR gene editing technology continues to expand globally, there is an increasing need for a workforce that is both knowledgeable about its theoretical foundations and trained in its practical use. While advanced, technology-driven STEM courses have the potential to improve student retention, they are often lecture-heavy and lack intentional engagement strategies that support deeper learning. Moreover, agriculture is the second most impacted sector by this technology, yet there is a significant lack of teaching materials focused on CRISPR in plant biology. To address these gaps, we developed a framework for teaching gene editing that incorporates multiple engagement strategies beyond traditional lecture-based instruction. This framework was implemented over two semesters in an Introduction to Gene Editing course at Tennessee State University, offered to both undergraduate and graduate students enrolled in a degree in Agricultural Sciences. This manuscript outlines the various strategies used in the course including active learning, multimodal instructional approaches and experiential learning strategies that can be adopted in other classrooms to effectively teach gene editing. Survey-based results from the course indicate a measurable increase in student comfort with designing and executing CRISPR-Cas based experiments. Societal Impact StatementPlant biology lacks accessible teaching materials for CRISPR, a powerful gene-editing technology widely used to improve agriculture. We developed an engaging framework to teach CRISPR concepts to advanced undergraduates and graduate students in plant sciences, which can be readily adopted by other instructors. The approach increased self-reported confidence in students and comfort in explaining CRISPR. Since instructors often have limited time to design interactive lessons, this framework offers a ready-to-use, effective strategy that makes CRISPR more widely available in classrooms, ultimately strengthening CRISPR literacy in the future agricultural workforce.

scientific communication and education↗

bioRxiv2025

Towards functional annotation with latent protein languagemodel features

Protein Language Models (PLMs) create high-dimensional embeddings that can be transformed into interpretable sparse features using Sparse Autoencoders (SAEs), where each feature activates on specific protein elements or patterns. However, scalably identifying which features are cohesive and reliable enough for protein annotation remains challenging. We address this by developing a validation pipeline combining three complementary methods: (1) expanded database matching across 20+ annotation sources including hierarchical codes, (2) feature-guided local structural alignment to identify structurally consistent activation regions, and (3) LLM-based feature description generation. Our annotation pipeline demonstrates three key properties of SAE features that make them a useful source of functional annotation complementary to existing methods. First, they can represent more granular patterns than existing protein databases, enabling the identification of sub-domains. Second, they can detect missing annotations by finding proteins that display recognizable structural motifs but lack corresponding database labels. Here, we automatically identify at least 491 missing CATH topology annotations with our pipeline. Third, they can maintain structural consistency across unseen proteins. Of our 10,240 SAE features, we find 615 that are consistently structurally similar in unannotated metagenomic proteins, allowing us to structurally match at least 8,077 metagenomic proteins to characterized proteins. This provides a rapid annotation pipeline with constant time search regardless of database size, that automatically includes structural and function information about the feature that triggered the match. Code is available at https://github.com/jsilbergDS/towards_functional_annotations_plms

bioinformatics↗