Biology subjects

Zhong, H.

Publications and source records attributed to Zhong, H..

At least 37 records · Page 2Linked to original sources

bioRxiv2025

A cross model spatial and single-cell atlas reveals the conserved involvement of osteopontin in polycystic kidney disease

Polycystic kidney disease (PKD) arises from mutations in cilia-associated genes, such as PKD1 and PKD2, expressed in renal epithelial cells, leading to progressive kidney dysfunction and end- stage kidney disease (ESKD). PKD patients exhibit significant heterogeneity in disease progression, largely due to genetic and environmental modifiers. Like patients, mouse models of PKD also exhibit significant heterogeneity with regards to the gene mutated, age of disease onset, and rate of disease progression. To elucidate the cellular and molecular consequences of these variables, we constructed an integrated single-cell and spatial transcriptomics atlas across mouse models of PKD, mapping changes in cell type composition, gene expression, and intercellular signaling networks across the whole atlas and within individual models. Consistently across models, single cell RNA sequencing (scRNAseq) data revealed increased Spp1 (osteopontin) expression and signaling from PKD-enriched clusters to Ly6clo monocytes. Global deletion of Spp1 in Pkd1RC/RC mice resulted in reduced cyst severity, improved kidney function, and reduced Ly6clo monocyte numbers, suggesting that SPP1 signaling to Ly6clo monocytes promotes PKD progression. We also created a freely available, searchable website (https://bmblx.bmi.osumc.edu/scPKD/) that can be used to identify cross- and intra-model specific changes in gene expression, guiding researchers to new therapeutic targets for treating PKD.

bioinformatics↗

bioRxiv2025

Unify: Learning Cellular Evolution with Universal Multimodal Embeddings

Integrating single-cell RNA-sequencing (scRNA-seq) data across species is hindered by evolutionary divergence, technical batch effects, and the reliance on one-to-one orthologs. We present Unify, a transfer learning methodology that learns universal cell embeddings by defining functionally coherent, multi-modal macrogenes. This is achieved by combining RNA expression with embeddings from protein language models and general-purpose language models. Unify transcends species boundaries, enabling cross-species comparisons beyond strict gene-level homology. Unify corrects batch effects while preserving conserved biological signals across vast evolutionary distances and enables more accurate prediction of perturbation responses across species, such as from mouse to human. Applied to species separated by over 700 million years, Unify reconstructs more accurate multi-species cell-type evolutionary trees and uncovers convergent gene programs. Together, these results establish Unify as a powerful method for comparative single-cell genomics and evolutionary biology.

bioinformatics↗

bioRxiv2025

Quantitative Proteomics Links Mitochondrial Dysfunction to Metabolic Changes and Epithelial Differentiation Defects in Hyperoxia-Exposed Neonatal Airway Cells

Premature infants often require supplemental oxygen therapy, however, exposure to supraphysiological oxygen (hyperoxia) can disrupt normal lung development and contribute to bronchopulmonary dysplasia (BPD). Mitochondrial dysfunction is increasingly recognized as a contributor to hyperoxia-induced BPD. However, the effects of hyperoxia on mitochondrial function and mucociliary differentiation in the developing upper airway epithelium remain poorly understood. This study tested the hypothesis that hyperoxia impairs neonatal airway mucociliary differentiation by disrupting mitochondrial bioenergetic function. Neonatal tracheal airway epithelial cells (nTAECs) from term infants (n=5) were cultured in a 3D air-liquid interface (ALI) model and exposed to 60% O2 during the mid-phase of differentiation (ALI day 7-14). Cellular phenotype was assessed using immunofluorescence staining and gene expression analyses. Mitochondrial function was evaluated through Seahorse metabolic flux analysis, and global protein changes were characterized by quantitative proteomics. Hyperoxia exposure significantly impaired terminal epithelial differentiation, characterized by reductions in ciliated and goblet cells. Seahorse assay revealed a decrease in baseline oxygen consumption and mitochondrial ATP production, accompanied by a compensatory increase in glycolytic ATP production. Quantitative proteomics identified disruption of mitochondrial Complex I as a central feature of the hyperoxic response. Downstream proteomic pathway analyses further confirmed the metabolic shift from mitochondrial to glycolytic ATP production and demonstrated altered epithelial differentiation pathways, including NOTCH and TGF-{beta} signaling. These findings reveal that moderate hyperoxia impairs mitochondrial bioenergetics and alters metabolic programming, leading to disrupted mucociliary differentiation. Future in vivo studies should evaluate mitochondrial oxidative fitness as a therapeutic target in neonatal lung disease. NEW & NOTEWORTHYWe report that moderate hyperoxia during a critical window of mucociliary differentiation disrupts terminal maturation in neonatal airway epithelial cells cultured in a 3D model. Hyperoxia induced mitochondrial bioenergetic dysfunction and metabolic reprogramming, with proteomic analysis identifying Complex I disruption as a key driver of impaired differentiation. Overall, these findings reveal a previously underrecognized link between mitochondrial bioenergetics and airway epithelial development, positioning metabolic dysfunction as an early trigger of hyperoxia-induced neonatal airway injury.

developmental biology↗

bioRxiv2025

Predictive Coding Explains Asymmetric Connectivity in the Brain: A Neural Network Study

Seminal frameworks of predictive coding propose a hierarchy of generative modules, each attempting to infer the neural representation of the module one level below; the predictions are carried by top-down feedback projections, while the predictive error is propagated by reciprocal forward pathways. Such symmetric feedback connections support visual processing of noisy stimuli in computational models. However, neurophysiological studies have yielded evidence of asymmetric cortical feedback connections. We investigated the contribution of neural feedback during sensorimotor processes, in particular visual processing during grasp planning, by utilizing convolutional neural network models that had been augmented with predictive feedback and were trained to compute grasp positions for real-world objects. After establishing an ameliorative effect of symmetric feedback on grasp detection performance when evaluated on noisy stimuli, we characterized the performance effects of asymmetric feedback, similar to that observed in the cortex. Specifically, we tested model variants extended with short-, medium- and long-range feedback connections (i) originating at the same source layer or (ii) terminating at the same target layer. We found that the performance-enhancing effect of predictive coding under adverse conditions was optimal for medium-range asymmetric feedback. Moreover, this effect was most prominent when medium-range feedback originated at a level of representational abstraction that was proximal to the input layer, in contrast to more distal layers. To conclude, our simulations show that introducing biologically realistic asymmetric predictive feedback improves model robustness to noisy visual stimuli in a neural network model optimized for grasp detection. Significance statementIt is commonly held that the brain predicts the causes of its sensorium via top-down neural pathways. While canonical models of predictive coding assume reciprocal feedforward and feedback connections, functional evidence highlights the importance of non-reciprocal asymmetric feedback, whose role remains poorly understood, particularly in sensorimotor functions. Using neural network models of grasp planning, we characterized optimal pathlengths and source regions for asymmetric feedback facilitating visuomotor processing of noisy sensory inputs. Our findings show that medium-range feedback from early layers marks a sweet spot, incorporating optimal distance between the neural representations of source/target layers and representational abstraction of the feedback source. This intimates an uncharted role of intermediate brain areas along the visuomotor stream as a source of predictive signals.

neuroscience↗

bioRxiv2025

Cerebellar climbing fibers convey perceptual choice during decision-making

Cerebellar climbing fibers are thought to signal reward prediction errors in non-motor functions. By imaging postsynaptic responses of climbing fibers onto mouse Purkinje cell dendrites during auditory discrimination, we found that climbing fibers in crus I can encode cue identities or perceptual choices. These responses were reshaped by reversal learning. Optogenetic perturbation of climbing fiber activity impaired discrimination. These results suggest a feedforward role of climbing fibers in perceptual decision-making.

neuroscience↗

bioRxiv2024

PhosF3C: A Feature Fusion Architecture with Fine-Tuned Protein Language Model and Conformer for prediction of general phosphorylation site

Protein phosphorylation, a key post-translational modification (PTM), provides essential insight into protein properties, making its prediction highly significant. Using the emerging capabilities of large language models (LLMs), we apply LoRA fine-tuning to ESM2, a powerful protein large language model, to efficiently extract features with minimal computational resources, optimizing task-specific text alignment. Additionally, we integrate the conformer architecture with the Feature Coupling Unit (FCU) to enhance local and global feature exchange, further improving prediction accuracy. Our model achieves state-of-the-art (SOTA) performance, obtaining AUC scores of 79.5%, 76.3%, and 71.4% at the S, T, and Y sites of the general data sets. Based on the powerful feature extraction capabilities of LLMs, we conduct a series of analyses on protein representations, including studies on their structure, sequence, and various chemical properties (such as Hydrophobicity (GRAVY), Surface Charge, and Isoelectric Point). We propose a test method called Linear Regression Tomography (LRT) which is a top-down method using representation to explore the models feature extraction capabilities, offering a pathway to improved interpretability.

bioinformatics↗

bioRxiv2024

At-RS31 orchestrates hierarchical cross-regulation of splicing factors and integrates alternative splicing with TOR-ABA pathways

O_LIAlternative splicing is essential for plants, enabling a single gene to produce multiple transcript variants to boost functional diversity and fine-tune responses to environmental and developmental cues. At-RS31, a plant-specific splicing factor in the Serine/Arginine (SR)-rich protein family, responds to light and the Target of Rapamycin (TOR) signaling pathway, yet its downstream targets and regulatory impact remain unknown. C_LIO_LITo identify At-RS31 targets, we applied individual-nucleotide resolution crosslinking and immunoprecipitation (iCLIP) and RNAcompete assays. Transcriptomic analyses of At-RS31 mutant and overexpressing plants further revealed its effects on alternative splicing. C_LIO_LIiCLIP identified 4,034 At-RS31 binding sites across 1,421 genes, enriched in CU-rich and CAGA RNA motifs. Comparative iCLIP and RNAcompete data indicate that the RS domain of At-RS31 may influence its binding specificity in planta, underscoring the value of combining in vivo and in vitro approaches. Transcriptomic analysis showed that At-RS31 modulates diverse splicing events, particularly intron retention and exitron splicing, and influences other splicing modulators, acting as a hierarchical regulator. C_LIO_LIBy regulating stress-response genes and genes in both TOR and abscisic acid (ABA) signaling pathways, At-RS31 may help integrate these signals, balancing plant growth with environmental adaptability through alternative splicing. C_LI

plant biology↗

bioRxiv2024

Genomic diversity and underlying metabolic variations of the human gut microbiome between populations at single nucleotide resolution

BackgroundA comprehensive and representative reference database is crucial for accurate taxonomic and functional profiling of the human gut microbiome in population-level studies. However, with 70% of current microbial reference data derived from Europeans and the Americans, East Asia, especially China, remain underrepresented. MethodsWe constructed the human Gut Microbiome Reference (GMR), comprising 478,588 high-quality microbial genomes from Chinese (247,134) and non-Chinese (231,454) populations. Species-level clustering and protein annotations were performed to characterize microbial diversity and function. We further integrated novel genomes into taxonomic profiling database and validated the improvements using independent cohort data. ResultsThe GMR dataset spans 6,664 species, including 26.4% newly classified species, and encodes over 20 million unique proteins, with 47% lacking known functional annotations. Notably, we observed that 35.35% and 32.46% of species unique to Chinese and non-Chinese populations, respectively. For 2,145 species shared between populations, 74% of 304 species with balanced prevalence between populations exhibited population-specific phylogenetic stratification, involving health relevant functionalities such as antibiotic resistance. Integration of novel genomes into taxonomic improved population-level species profiling by up to 23% and uncovered replicable associations between novel species and host physiological traits. ConclusionsOur study largely expands the compositional and functional landscape of the human gut microbiome, providing a crucial resource for studying the role of gut microbiome for regional health disparities.

genomics↗

bioRxiv2024

Haplotype-resolved assemblies provide insights into genomic makeup of the oldest grapevine cultivar (Munage) in Xingjiang

AbstractsMunage, an ancient grape variety that has been cultivated for thousands of years in Xinjiang, China, is recognized for its exceptional fruit traits. There are two main types of Munage: white fruit (WM) and red fruit (RM). However, the lack of a high-quality genomic resources has impeded effective breeding and restricted the potential for expanding these varieties to other growing regions. In this study, we assembled haplotype-resolved genome assemblies for WM and RM, alongside integrated whole genome resequencing (WGS) data and transcriptome data to illuminate specific mutations and associated genes in Munake and the genes associated with fruit color traits. Selective analysis between Munage clones and Eurasian grapes suggested that adaptive selection exists in Munage grapes, with genes enriched in processes including cell maturation, plant epidermal cell differentiation, and root epidermal cell differentiation. The study examined the mutations within Munage grapes and found that the genes PMAT2 on chromosome 12 and MYB123 on chromosome 13 are likely responsible for color variation in RM. These findings provide crucial genetic resources for investigating the genetics of the ancient Chinese grape variety, Munage, and will facilitate the genetic improvement in grapevine.

genomics↗

bioRxiv2024

ER stress-induced TREM2 downregulation exacerbates platelet activation and myocardial infarction in patients with coronary artery disease

Coronary artery disease (CAD) is characterized by the chronic immune-inflammation, excessive endoplasmic reticulum (ER) stress, and platelet hyperactivity; however, whether there is a signaling hub linking these events remains unclear. Here, we identified that triggering receptor expressed on myeloid cells 2 (TREM2), an important pattern recognition receptor of the innate immune system, may serve as one such hub. We found that platelets expressed TREM2 and platelets from CAD patients had decreased TREM2 expression compared to healthy subjects. Decreased TREM2 is associated with platelet hyperactivity in CAD patients. This decrease could be due to excessive ER stress, which downregulated TREM2 through the CHOP-C/EBP axis. Loss of TREM2 not only enhanced platelet activation in response to ADP, collagen, and collagen-related peptide (CRP), but also amplified the platelet inflammatory response. Loss of TREM2 exacerbated mouse mesenteric arterial thrombosis and aggravated experimental myocardial infarction (MI). Moreover, a TREM2-activating antibody inhibited platelet activation, alleviated arterial thrombosis and pulmonary embolism. In addition, TREM2-activating antibody exhibited cardioprotective roles against experimental MI and reduced the inflammatory burden. Mechanistically, TREM2/DAP12/SHIP1 axis negatively regulated platelet activation through reducing PIP3 levels and inhibiting Akt phosphorylation. We also provided evidence supporting sphingosine-1-phospage (S1P) as a physiological agonist of TREM2. In summary, we find that TREM2 connects chronic immune-inflammation, excessive ER stress, and platelet hyperactivity in CAD patients. Downregulating TREM2 by ER stress exacerbates platelet activation and amplifies inflammation response in patients with CAD. TREM2-activating antibodies may have therapeutic potential for CAD patients. Key PointsO_LIPlatelets from CAD patients have decreased TREM2 expression, which is caused by ER stress and associated with platelet hyperactivity. C_LIO_LIS1P-TREM2-SHIP1 pathway inhibits platelet activation, alleviates arterial thrombosis, and exhibits cardioprotective roles against experimental myocardial infarction. C_LI O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=88 SRC="FIGDIR/small/608887v1_ufig1.gif" ALT="Figure 1"> View larger version (26K): org.highwire.dtl.DTLVardef@1e21f2eorg.highwire.dtl.DTLVardef@6d427forg.highwire.dtl.DTLVardef@128af74org.highwire.dtl.DTLVardef@16f48c_HPS_FORMAT_FIGEXP M_FIG O_FLOATNOVisual Abstract.C_FLOATNO Downregulation of platelet TREM2 caused by ER stress in CAD patients leads to platelet hyperactivity and aggravates myocardial ischemia, which is rescued by TREM2-activating antibody. Excessive ER stress in CAD upregulates CHOP, which dimerizes with C/EBP, a transcription factor of TREM2, decreases TREM2 transcription and expression in megakaryocytes and further in platelets. TREM2/DAP12/SHIP1/Akt pathway negatively regulates platelet activation; decreased TREM2 thus aggravates platelet hyperactivity, inflammation, and myocardial infarction, which can be rescued by TREM2 activation including antibody or the small molecule agonist. Physiologically, S1P released from platelet granules is an endogenous agonist of TREM2. C_FIG

pharmacology and toxicology↗

bioRxiv2024

A high-performance genetically encoded sensor for cellular imaging of PKC activity in vivo

We report a genetically encoded fluorescence lifetime sensor for protein kinase C (PKC) activity, named CKAR3, based on Forster resonance energy transfer. CKAR3 exhibits a 10-fold increased dynamic range compared to its parental sensors and enables in vivo imaging of PKC activity during animal behavior. Our results reveal robust PKC activity in a sparse neuronal subset in the motor cortex during locomotion, in part mediated by muscarinic acetylcholine receptors.

neuroscience↗

bioRxiv2024

Cryo-electron tomographic investigation of native hippocampal glutamatergic synapses

Chemical synapses are the major sites of communication between neurons in the nervous system and mediate either excitatory or inhibitory signaling [1]. At excitatory synapses, glutamate is the primary neurotransmitter and upon release from presynaptic vesicles, is detected by postsynaptic glutamate receptors, which include ionotropic AMPA and NMDA receptors [2, 3]. Here we have developed methods to identify glutamatergic synapses in brain tissue slices, label AMPA receptors with small gold nanoparticles (AuNPs), and prepare lamella for cryo-electron tomography studies. The targeted imaging of glutamatergic synapses in the lamella is facilitated by fluorescent pre- and postsynaptic signatures, and the subsequent tomograms allow for identification of key features of chemical synapses, including synaptic vesicles, the synaptic cleft and AuNP-labeled AMPA receptors. These methods pave the way for imaging brain regions at high resolution, using unstained, unfixed samples preserved under near-native conditions.

biochemistry↗

bioRxiv2024

Prevalent co-release of glutamate and GABA throughout the mouse brain

Several neuronal populations in the brain transmit both the excitatory and inhibitory neurotransmitters, glutamate, and GABA, to downstream neurons. However, it remains largely unknown whether these opposing neurotransmitters are co-released onto the same postsynaptic neuron simultaneously or are independently transmitted at different time and locations (called co-transmission). Here, using whole-cell patch-clamp recording on acute mouse brain slices, we observed biphasic miniature postsynaptic currents, i.e., minis with time-locked excitatory and inhibitory currents, in striatal spiny projection neurons (SPNs). This observation cannot be explained by accidental coincidence of monophasic miniature excitatory and inhibitory postsynaptic currents (mEPSCs and mIPSCs, respectively), arguing for the co-release of glutamate and GABA. Interestingly, these biphasic minis could either be an mEPSC leading an mIPSC or vice versa. Although dopaminergic axons release both glutamate and GABA in the striatum, deletion of dopamine neurons did not eliminate biphasic minis, indicating that the co-release originates from another neuronal type. Importantly, we found that both types of biphasic minis were detected in other neuronal subtypes in the striatum as well as in nine out of ten additionally tested brain regions. Our results suggest that co-release of glutamate and GABA is a prevalent mode of neurotransmission in the brain.

neuroscience↗

bioRxiv2024

Emergence of Orientation Pinwheels in a Self-Evolving Spiking Neural Network: Enhancing Visual Coding Efficiency and Reliability

Orientation preference maps (OPMs) in the primary visual cortex of primates organize orientation-tuned neurons into columnar structures, forming pinwheel-like patterns. However, lower-level animals like rodents typically exhibit a lack of OPMs, with neurons either randomly distributed or aggregated in small clusters. This distinction prompts an inquiry into whether more structured cortical columns correlate with improved visual computational or coding efficiency. To explore this, we propose a novel self-evolving spiking neural network (SESNN). To the best of our knowledge, the SESNN is the first spiking network, incorporating mechanisms of neural plasticity in forming neural connections without explicit objective functions. We reveal that the emergence of pinwheel structures is primarily driven by sparse coding constraints and local synaptic plasticity as fundamental mechanisms. Second, for higher mammals with expansive iso-orientation domains (IODs), the firing responses in pinwheel structures primarily emanate from pinwheel centers (PCs) and progressively extend toward the periphery, encompassing adjacent IODs. Third, the size and organization of these IODs across species are significantly influenced by the receptive fields ability to process overlapping visual information. Lastly, PCs within large IODs demonstrate enhanced robustness and population sparseness in detecting a variety of orientation features. These results indicate that the spatial pinwheel structure facilitates highly efficient and reliable coding performance.

neuroscience↗

bioRxiv2024

The Functional Role of Pinwheel Topology in the Primary Visual Cortex of High-Order Animals for Complex Natural Image Representation

The primary visual cortex (V1) of high-level animals exhibits a complex organization of neuronal orientation preferences, characterized by pinwheel structure topology, yet the functional role of those complex patterns in natural image representation remains largely unexplored. Our study first establishes a new self-evolving spiking neural network (SESNN) model, designed to mimic the functional topological structure of orientation selectivity within V1. We observe the emergence of a particularly new "spread-out" firing patterns from center to the surround of the pinwheel structures in response to natural visual stimuli in pinwheel structures, propagating from pinwheel centers and spreading to iso-orientation domains--a pattern not found in salt- and-pepper organizations. To investigate this phenomenon, we propose a novel deep recurrent U-Net architecture to reconstruct images from V1s spiking activity across time steps and assess the encoded information entropy of different firing patterns via the models predicted uncertainty, offering a spatiotemporal analysis of V1s functional structures. Our findings reveal a trade-off between visual acuity and coding time: the "spread-out" pattern enhances the representation of complex visual details at the cost of increased response latency, while salt-and-pepper organizations, lacking such domains, prioritize rapid processing at the expense of reduced visual acuity. Additionally, we demonstrate that this trade-off is modulated by the size of iso-orientation domains, with larger domains--supported by denser neuronal populations--substantially improving both visual acuity, coding efficiency, and robustness, features diminished in smaller domains and salt-and-pepper arrangements. Our research provides a foundational understanding of the principles underlying efficient visual information representation and suggests novel strategies for advancing the robustness and performance of image recognition algorithms in artificial intelligence.

neuroscience↗

bioRxiv2024

A 10-valent composite mRNA vaccine against both influenza and COVID-19

The COVID-19 pandemic caused by SARS-CoV-2 viruses has had a persistent and significant impact on global public health for four years. Recently, there has been a resurgence of seasonal influenza transmission worldwide. The co-circulation of SARS-CoV-2 and seasonal influenza viruses results in a dual burden on communities. Additionally, the pandemic potential of zoonotic influenza viruses, such as avian Influenza A/H5N1 and A/H7N9, remains a concern. Therefore, a combined vaccine against all these respiratory diseases is in urgent need. mRNA vaccines, with their superior efficacy, speed in development, flexibility, and cost-effectiveness, offer a promising solution for such infectious diseases and potential future pandemics. In this study, we present FLUCOV-10, a novel 10-valent mRNA vaccine created from our proven platform. This vaccine encodes hemagglutinin (HA) proteins from four seasonal influenza viruses and two avian influenza viruses with pandemic potential, as well as spike proteins from four SARS-CoV-2 variants. A two-dose immunization with the FLUCOV-10 elicited robust immune responses in mice, producing IgG antibodies, neutralizing antibodies, and antigen-specific cellular immune responses against all the vaccine-matched viruses of influenza and SARS-CoV-2. Remarkably, the FLUCOV-10 immunization provided complete protection in mouse models against both homologous and heterologous strains of influenza and SARS-CoV-2. These results highlight the potential of FLUCOV-10 as an effective vaccine candidate for the prevention of influenza and COVID-19. Author summaryAmidst the ongoing and emerging respiratory viral threats, particularly the concurrent and sequential spread of SARS-CoV-2 and influenza, our research introduces FLUCOV-10. This novel mRNA-based combination vaccine, designed to counteract both influenza and COVID-19, by incorporating genes for surface glycoproteins from various influenza viruses and SARS-CoV-2 variants. This combination vaccine showed highly effective in preclinical trials, generating strong immune responses, and ensuring protection against both matching and heterologous strains of influenza and SARS-CoV-2. FLUCOV-10 represents a significant step forward in our ability to address respiratory viral threats, showcasing potential as a singular, adaptable vaccine solution for global health challenges.

microbiology↗

bioRxiv2024

OligoFormer: an accurate and robust prediction method for siRNA design

MotivationRNA interference(RNAi) has become a widely used experimental approach for post-transcriptional regulation and is increasingly showing its potential as future targeted drugs. However, the prediction of highly efficient siRNAs(small interfering RNA) is still hindered by dataset biases, the inadequacy of prediction methods, and the presence of off-target effects. To overcome these limitations, we propose an accurate and robust prediction method, OligoFormer, for siRNA design. ResultsOligoFormer comprises three different modules including thermodynamic calculation, RNA-FM module, and Oligo encoder. Oligo encoder is the core module based on the transformer encoder. Taking siRNA and mRNA sequences as input, OligoFormer can obtain thermodynamic parameters, RNA-FM embedding, and Oligo embedding through these three modules, respectively. We carefully benchmarked OligoFormer against 5 comparable methods on siRNA efficacy datasets. OligoFormer outperforms all the other methods, with an average improvement of 9% in AUC and 10.7% in F1 score in our inter-dataset validation. We also provide a comprehensive pipeline with prediction of siRNA efficacy and off-target effects using PITA score and TargetScan score. The ablation study shows RNA-FM module and thermodynamic parameters improved the performance and accelerated convergence of OligoFormer. The saliency map by gradient backpropagation shows certain base preferences in initial and terminal region of siRNAs.

bioinformatics↗

bioRxiv2023

Integrative genomics reveals the polygenic basis of seedlessness in grapevine

Seedlessness is a crucial quality trait in table grape (Vitis vinifera L.) breeding. However, the development of seeds involved intricate regulations, while the polygenic basis of seed abortion remains unclear. Here, we combine comparative genomics, population genetics, quantitative genetics, and integrative genomics to unravel the evolution and polygenic basis of seedlessness in grapes. We generated four haplotype-resolved telomere-to-telomere (T2T) genomes for two seedless grape cultivars, Thompson Seedless (TS, syn. Sultania) and Black Monukka (BM). Comparative genomics identified a [~]4.25 Mb hemizygous inversion on Chr10 specific in seedless cultivars, with seedless-associated genes VvTT16 and VvSUS2 located at breakpoints. Population genomic analyses of 548 grapevine accessions revealed two distinct clusters of seedless cultivars, tracing the origin of the seedlessness trait back to Sultania. Introgression, rather than convergent selection, shaped the evolutionary history of seedlessness in grape improvement. Genome-wide association study (GWAS) analysis identified 110 quantitative trait loci (QTLs) associated with 634 candidate genes, including novel candidate genes, such as three 11S GLOBULIN SEED STORAGE PROTEIN and two CYTOCHROME P450 genes, and well-known genes like VviAGL11. Integrative genomic analyses resulted in 339 core candidate genes categorized into 13 groups related to seed development. Machine learning based genomic selection achieved a remarkable 99% precision in predicting grapevine seedlessness. Our findings highlight the polygenic nature of seedless and provide novel candidate genes for molecular genetics and an effective prediction for seedlessness in grape genomic breeding.

genetics↗