Biology subjects

Yu, H.

Publications and source records attributed to Yu, H..

At least 91 records · Page 5Linked to original sources

bioRxiv2026

Constructing the ensemble of representative structures for a protein via neural-surrogate-guided MSA recombination

AO_SCPLOWBSTRACTC_SCPLOWStructural dynamics is essential for the functional and mechanistic illustration of proteins. Previous research attempted to generate diversified protein structures by utilizing the multiple sequence alignment (MSA), but failed to provide physically relevant representative conformations without state annotations. In this work, we propose a framework named ProCEDiS to generate a compact ensemble of representative conformations for the target protein without prior knowledge. Adopting a neural surrogate to assist the exploration of MSA recombination and integrating with AlphaFold2 to model structures, this method can automatically find high-quality, mutually dissimilar conformations for the target sequence. Parallel short-timescale molecular dynamics (MD) simulations on these structure seeds enable quick while crude free energy estimation, from which physically plausible representative states could be identified. In the benchmark on four protein systems, the ProCEDiS + MD pipeline is capable of providing valuable structural dynamics information within acceptable running time.

bioinformatics↗

bioRxiv2026

The novel variant IBDV induces inflammatory responses via VP3-mediated pyroptosis by directly targeting Caspase-9

The novel variant infectious bursal disease virus (nVarIBDV), which has been widely prevalent since 2017, threatens the poultry industry by inducing severe bursal atrophy and intense inflammatory responses. Understanding the inflammatory mechanism underlying nVarIBDV infection is critical for preventing virus-induced damage. Pyroptosis, an inflammatory type of programmed cell death, may contribute to the inflammation and bursa of Fabricius damage induced by nVarIBDV. Here, we found that the nVarIBDV infection induced severe inflammatory responses both in vivo and in vitro, which were associated with pyroptosis. Further research revealed that the viral protein VP3 drove pyroptosis induced by nVarIBDV infection by indirectly activating the Caspase-3-GSDME pathway, leading to GSDME cleavage. Mechanistically, VP3 directly interacted with and activated Caspase-9, thereby initiating Caspase-9-Caspase-3-GSDME pathway. More importantly, the Serine 33 of VP3 was identified as the key amino acid for interacting with and activating Caspase-9. Mutation of this residue significantly weakened the ability of VP3 to interact with Caspase-9 and activate the Caspase-9-mediated pyroptosis pathway, altered the binding mode of the Caspase-9-VP3 protein complex, and ultimately reduced the ability of VP3 to induce pyroptosis. In conclusion, our results elucidated a novel mechanism by which nVarIBDV infection induced inflammatory responses whereby viral protein VP3 triggered pyroptosis by targeting the Caspase-9-Caspase-3-GSDME pathway. IMPORTANCEnVarIBDV is globally prevalent and poses a significant threat to the poultry industry by causing inflammation and damage to the bursa of Fabricius. This study revealed that nVarIBDV infection induced inflammatory responses and bursa of Fabricius damage, primarily mediated by the viral protein VP3 targeting Caspase-9 to induce pyroptosis. To our knowledge, this is the first report on the interaction between inflammatory responses induced by IBDV infection and pyroptosis. In addition, serine 33 was identified as the critical amino acid required for maintaining the stability of the interaction between VP3 and Caspase-9 and activation of the Caspase-9-Caspase-3-GSDME cascade pyroptosis pathway and further inflammatory responses. Collectively, our findings not only elucidate a novel mechanism by which nVarIBDV induces pyroptosis and further leads to inflammation but also suggest that the identification of this serine site may represent a potential therapeutic target to block nVarIBDV-induced inflammation and bursa of Fabricius damage.

microbiology↗

bioRxiv2026

A plug-and-play ROI imaging module and deep-learning denoising framework extend three-photon microscopy to 1.7 mm depth

Three-photon microscopy (3PM) has extended optical imaging beyond the reach of two-photon microscopy (2PM), but its practical depth remains far below theoretical predictions because of low photon efficiency and severe noise contamination. Here we present an integrated hardware and software solution that addresses these limitations. We developed a plug-and-play region-of-interest (ROI) imaging module that selectively excites neuron-occupied regions to improve power efficiency and reduce photothermal stress. The module incorporates frame-partitioned accumulative (FPA) imaging for high signal-to-noise ratio (SNR) structural acquisition, automated neuron segmentation, and motion-robust registration for stable recordings. Complementing this, OptiCal is a cascaded deep-learning framework that removes mixed noise including periodic, motion-induced, and random components. Together, these innovations enable high-fidelity imaging to 1.7 mm depth, extending the practical limit of 3PM by about 400 {micro}m while maintaining low excitation power. Our results reveal reliable calcium activity and behavior-correlated dynamics in deep medial prefrontal cortex of awake mice.

neuroscience↗

bioRxiv2025

Kilohertz-rate two-photon voltage imaging of population dynamics in vivo

Understanding how neural circuits compute requires capturing voltage dynamics across large neuronal populations with millisecond resolution in vivo. However, two-photon voltage imaging remains fundamentally limited by a trade-off among imaging speed, field of view, and excitation efficiency. We introduce HS2PM, a hybrid scanning two-photon microscope that overcomes this bottleneck, enabling kilohertz-rate imaging over a 650 x 524 m2 field of view at single-cell resolution while preserving the photon efficiency of single-point excitation. HS2PM stably records deep-layer membrane potentials from over 160 neurons simultaneously, with minimal photobleaching and phototoxicity. It resolves both spikes and subthreshold voltage dynamics in vivo, revealing how these jointly shape sensory adaptation and population coding. Beyond voltage imaging, HS2PM supports high-speed fluorescence lifetime and vascular flow imaging, establishing a multimodal platform for dissecting fast circuit dynamics with precision previously inaccessible to optical methods.

neuroscience↗

bioRxiv2025

Assembly and glycosylation of Helicobacter pylori sheathed flagella

The bacterial flagellum is a complex nanomachine essential for motility, colonization, and invasion in diverse species. Helicobacter pylori has evolved elaborate sheathed flagella that enable migration through the highly viscous gastric mucus layer to reach its colonization niche on the gastric epithelium, yet the molecular basis for these unique adaptations has remained elusive. Here, we use in-situ single particle cryo-electron microscopy to determine near-atomic structures of the flagellar filament within the membranous sheath of H. pylori. The major flagellin FlaA constitutes the bulk of the filament, whereas the minor flagellin FlaB contributes critically to the hook-proximal region. Both FlaA and FlaB form a conserved core surrounded by variable surface-exposed domains. Our structures further reveal that pseudaminic-acid glycans decorate these domains, where they mediate inter- and intra-subunit contacts that stabilize the filament and confer a negatively charged surface. Together, these findings support a model in which the filament rotates independently of the membranous sheath to drive H. pylori motility and provide a molecular framework for understanding how the sheathed flagellum enables colonization and persistence within the gastric niche. Significance StatementWe present the first in-situ near-atomic structure of the sheathed flagellar filament in Helicobacter pylori, revealing distinctive adaptations that underpin the pathogens unique motility and persistent infection. Our in-situ structures show that the two flagellins, FlaA and FlaB, assemble into an extended and exceptionally stable filament through an extensive hydrogen-bonding network. Pseudaminic acid glycans decorate the surface-exposed domains, where they stabilize inter-subunit packing and render the surface negatively charged and hydrophilic. These findings, which provide insight into the assembly of the flagellar filament and its relationship to the surrounding sheath, provide a structural framework for developing strategies to disrupt H. pylori motility and infection.

microbiology↗

bioRxiv2025

Human trunk embryoids with patterned anterior-posterior and dorsal-ventral body axes: utility for understanding human development and disease

Human embryoid models enable mechanistic studies of development and disease. We generated trunk embryoids from human pluripotent stem cells that recapitulate posterior trunk formation at Carnegie stage (CS) 8-10, with patterned anterior-posterior (A-P) and dorsal-ventral (D-V) axes. These self-organizing structures comprise a ventral notochord, dorsal neural tube, floor plate and bilateral somites. Genetic and chemical perturbations of SHH signaling confirmed the notochords central role in D-V patterning. Moreover, VANGL1/2 loss-of-function mutations recapitulated mouse phenotypes, including axial truncation and somite segmentation failure. This model enables detailed study of key developmental events that underlie posterior trunk formation and provides a promising platform for human disease modeling.

developmental biology↗

bioRxiv2025

Refining liquid chromatography conditions across platforms to democratize single-cell proteomics

Single-cell proteomics has emerged as a powerful approach for characterizing cellular heterogeneity. Here, we present an optimized scProteomic workflow that enhances proteome coverage and quantification by refining liquid chromatography (LC) conditions across platforms (both Evosep and nanoElute2). Importantly, our results show that using our optimized LC conditions, even with Bruker timsTOF HT, a machine not designed for scProteomics applications, we achieved solid performance with single cell samples that allows meaningful biological discoveries. First, we compared power to detect differentially-expressed genes/proteins between scRNA-seq and scProteomics at single cell level, and demonstrated that scProteomics showed an excellent performance at small sample sizes. To further validate this finding, we applied our optimized workflow to study the proteomic response to oxidative stress using a limited number of single cells. Our scProteomics results successfully detected 3 distinct cell populations, reflecting correctly the 3 cell lines used, and captured the dysregulation of antioxidant enzymes, molecular chaperones and ubiquitin-proteasome system, reflecting a multi-faceted response to oxidative stress that is uniform across distinct cell lines. Our results highlighted the potential of scProteomics to resolve subtle perturbations and provide a readout of cellular states for the broader community, including users operating non-scProteomics-dedicated machines.

systems biology↗

bioRxiv2025

Proteostasis Remodeling Across Development Defines Fetal, Neonatal, and Adult Hematopoietic Stem Cell States

Hematopoietic stem cells (HSCs) must preserve protein homeostasis (proteostasis) despite dramatic changes in proliferative and biosynthetic demands during development, yet how proteostasis is regulated across these transitions is poorly understood. Here, we show that fetal and neonatal HSCs operate through distinct, stage-specific proteostasis programs that differ fundamentally from those in adulthood. Using quantitative in vivo assays spanning embryonic through adult stages, we uncover an unanticipated decoupling between protein synthesis and protein quality control during development, revealing that fetal and neonatal HSCs employ specialized mechanisms to safeguard proteome integrity under developmental stress. Developing HSCs experience a distinctive proteostasis landscape characterized by elevated protein synthesis, increased unfolded protein burden, and selective engagement of stress-buffering and protein degradation pathways that are largely dispensable in young adult HSCs. Disruption of these pathways compromises early life HSC function and long-term fitness, establishing proteostasis control as a key regulator of stem cell maturation. These findings define previously unrecognized mechanisms by which HSCs manage the proteome during early life and reveal fundamental principles governing stem cell proteostasis across ontogeny.

developmental biology↗

bioRxiv2025

Synchronized Latency Reversal and Immune Clearance by a Multifunctional Fusion Protein Enables HIV-1 Reservoir Reduction

Strategies that concurrently reactivate latent reservoirs and enhance immune-mediated clearance hold significant promise for achieving an HIV cure. Here, we developed hyperIL-15xsCD4-Fc (15xsCD4-Fc), a fusion protein that integrates latency reactivation, targeted immune engagement, and effector-mediated killing into a single molecule. This agent not only potently reverses HIV-1 latency in CD4+ T cells from people living with HIV-1 (PLWH) through coordinated IL-15 receptor signaling and sCD4-mediated HIV-1 envelope (Env) engagement, but also enhances antigen-specific CD8+ T cell response in PBMCs derived from PLWH. Furthermore, 15xsCD4-Fc enables Env-specific elimination of reactivated latently infected cells by NK cells while preventing off-target cytotoxicity. In PBMCs from PLWH, 15xsCD4-Fc reduced replication-competent HIV-1 DNA by 93.8%. In antiretroviral treated HIV-1-infected humanized mice, the molecule demonstrated both safety and efficacy in diminishing viral reservoir in lymphoid organs. This spatiotemporally coupled approach to reservoir exposure and immune recognition establishes a clinically viable strategy for clearing the HIV-1 reservoir. Graphical abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=195 SRC="FIGDIR/small/693434v1_ufig1.gif" ALT="Figure 1"> View larger version (44K): org.highwire.dtl.DTLVardef@8f8f0org.highwire.dtl.DTLVardef@10f63f7org.highwire.dtl.DTLVardef@16c1de2org.highwire.dtl.DTLVardef@11e1787_HPS_FORMAT_FIGEXP M_FIG C_FIG

immunology↗

bioRxiv2025

Identification of the protein precursor for thyroid hormone synthesis in the basal chordate ascidian Styela clava

Thyroid hormones (THs) play crucial roles in development, growth, and metabolism across the animal kingdom1,2. Although the mechanism of TH synthesis in vertebrates has been well described, it remains elusive in invertebrates due to the lack of identified orthologs of the thyroglobulin (TG), the protein precursor for TH synthesis3,4. Here, we identified a functional TG protein encompassing multiple TG type 1 (Tg1)-domains, encoded by the ScTg-like gene in proto-vertebrate ascidian Styela clava based on the immunoprecipitation-mass spectrometry combined with phylogenetic and expression pattern analyses. In vitro iodination experiments demonstrated that ScTG-like has the capacity to provide hormonogenic sites for TH synthesis. Furthermore, in vivo gene knockdown of ScTg-like revealed that the larval metamorphosis was significantly inhibited as a consequence of diminished TH levels, indicating that ScTg-like is responsible for the production of endogenous TH. Additionally, an invaginated follicle-like structure was first discovered at the central anterior of the larval trunk where both ScTG-like protein and THs were deposited, suggesting that this structure serves as the location for THs synthesis and storage, functionally similar to vertebrate thyroid follicles. Moreover, we analyzed the structural features of ScTG-like protein and detected the similarly organized Tg1-containing proteins in multiple bilaterian phyla, suggesting that endogenous TH synthesis may be an ancestral, possibly synapomorphic, trait of bilaterians. Our study is the first to identify the TH protein precursor outside vertebrates by providing the direct evidence of endogenous de novo TH synthesis, which offers insights into the evolution of TG protein and the emergence of a major vertebrate endocrine system.

evolutionary biology↗

bioRxiv2025

Decoding Neoantigen-encoding Tumor-Specific Transcripts Unveils a Shared Target Reservoir for Immunotherapy in Hepatocellular Carcinoma

Background and AimsPrimary liver cancer, predominantly hepatocellular carcinoma (HCC), has limited therapeutic options. While mutation-derived neoantigen vaccine holds promise, its success is hindered by low antigen availability. This study explores transcriptome-derived neoantigens (neoantigen-encoding tumor-specific transcripts, neoTSTs) in HCC, characterizing their features, generation mechanisms, and therapeutic potential. Approach and ResultsWe analyzed RNA-seq data from 1,013 liver cancer patients and constructed a multi-layered reference dataset. Using a customized pipeline, we identified an average of 60 neoTSTs per patient, significantly surpassing mutation-derived neoantigens (neoMuts). NeoTSTs exhibited higher population frequencies, with 73.1% providing multiple epitopes, and were validated through mass spectrometry and HLA transgenic mouse models. Mechanistically, neoTSTs were generated via retained introns, transposable element activation, HNF4A-regulated alternative promoters, and de novo transmembrane domain (TMD) generation. Single-cell analysis revealed neoTSTs cover >75% of tumor cells and identified antigen-presenting cancer-associated fibroblasts (apCAFs) that enriched in immunotherapy responders and amplified CD4 T-cell responses via MHC-II presentation. In murine HCC models, neoTST vaccination outperformed neoMuts, inducing dual MHC-I/II activation and significant tumor growth inhibition. ConclusionsNeoTSTs represent a superior neoantigen source in HCC, compensating for the limitations of mutation-derived targets. Their abundance, sharedness, and dual MHC pathway activation highlight their potential for personalized immunotherapy, particularly in low-TMB tumors. Graphic Abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=178 SRC="FIGDIR/small/692877v1_ufig1.gif" ALT="Figure 1"> View larger version (44K): org.highwire.dtl.DTLVardef@1bedfedorg.highwire.dtl.DTLVardef@5f6c95org.highwire.dtl.DTLVardef@d1deb5org.highwire.dtl.DTLVardef@743e85_HPS_FORMAT_FIGEXP M_FIG C_FIG

cancer biology↗

bioRxiv2025

Acoustoluminescence in Transition Metal and Rare Earth Oxides Beyond 1800 nm for In Vivo Imaging

Acoustoluminescence (AL) is promising for molecular imaging in living tissue, but efficient acousto-optic conversion remains challenging due to the [~]108 times difference in quantum energy between phonons (1 MHz ultrasound) and photons (visible or near-infrared light). Here, we report AL in transition metal oxides (TMOs) and rare earth oxides (REOs) at wavelengths beyond 1800 nm in the short-wave infrared (SWIR) or near-infrared II (NIR-II, 1000-3000 nm) window, under ultrasound excitation at power densities 100-150-fold lower than those required for sonoluminescence in liquids. High-temperature N2/H2-mixed gas reduction was demonstrated as a safe and efficient method to regulate the AL spectra and brightness of TMOs and REOs. TMOs exhibited broadband NIR-II AL emission. Intrinsic emission peaks of rare earth ions and non-conventional luminescence were observed in the AL spectra of REOs under ultrasound excitation. NIR-II AL imaging enabled twice the penetration depth of fluorescence imaging. We developed a scanning focused ultrasound AL imaging system for in vivo tumor imaging through the intact hindlimb, achieving acoustic resolution and penetration depths exceeding one centimeter.

bioengineering↗

bioRxiv2025

Beyond Inheritance: De novo Fast Motion Computation in Primate Visual Cortex

Objects move through space and time, generating sequential visuotopic activations in all sighted animals leading to motion perception of velocity defined by direction and speed. Humans can effortlessly see motion with speeds ranging from 0.25 to 500{degrees}/s. However, direction-selective neurons in the primary visual cortex (V1)--from which all subsequent processing is presumed to derive-- only encode directionality at low speeds. To resolve this paradox, we recorded neuronal responses to moving dots, gratings, and movies across the LGN, V1, MT and MST of the macaque motion pathway. Regardless of cell type and motion stimuli, V1 neurons lost direction selectivity at [~]29{degrees}/s while MT and MST neurons maintained it up to [~]82{degrees}/s and [~]183{degrees}/s, respectively. A cascaded spatiotemporal integration model reveals that at each cortex direction-selective neurons can generate velocity selectivity de novo, by integrating sequential visuotopic activations from preceding areas, irrespective of speed and directionality. By computing velocity anew, the primate brain effectively uses the cortical hierarchy itself to shift gears to efficiently encode slow and fast motion. Thus, five visual areas from the retina into the brains processing hierarchy, external spatiotemporal information is being computed afresh, offering insights for motion processing in other species, modalities and machine vision.

neuroscience↗

bioRxiv2025

Near-atomic in-situ architecture and membrane-coupled dynamics of the Vibrio cholerae sheathed flagellum

The sheathed flagellum of Vibrio cholerae is a self-assembling membranous organelle that must coordinate axial assembly, sheath biogenesis, and rapid motor rotation. Here, we determine in-situ near-atomic structures of the sheathed flagellar motor inside intact cells. The motor anchors to the outer membrane through lipidated HL-rings without forming a membrane pore, thereby allowing axial assembly to drive sheath formation. Conserved LP-rings act as slide-rotary bushings that permit high-speed rotation within a dynamic envelope yet can constrict to seal the pore upon stress-induced ejection. We further show that stator activation requires a specific PomB-MotX interaction rather than peptidoglycan engagement. Together, these findings reveal how the distinctive architecture and dynamics of the sheathed flagellum promote V. cholerae motility, environmental survival, and persistent colonization of the human gut.

microbiology↗

bioRxiv2025

Accurate and robust 3D genome feature discovery from multiplexed DNA FISH

Multiplexed DNA FISH (also known as FISH Omics) is a powerful imaging technology to map three-dimensional (3D) genome organization in thousands of single cells at kilobase resolution. However, existing computational approaches to identify 3D genome features, such as A/B compartments, topologically associating domains (TADs) and chromatin loops, have been mainly adapted from Hi-C and do not account for the distinct measurement errors across the three XYZ physical axes. Here we present ArcFISH (accurate and robust chromatin analysis from FISH data), which explicitly models axis-specific measurement errors in a unified statistical framework to detect multi-level 3D genome features. Comprehensive evaluations using multiple simulated as well as experimental datasets show that ArcFISH reliably identifies chromatin loops under heterogeneous measurement errors, detects TAD boundaries highly enriched for active histone marks H3K4me3 and H3K36me3, and requires as few as 30 chromatin traces to faithfully assign A/B compartments. Applying ArcFISH to multiplexed DNA FISH data from mouse embryonic stem cells and mouse excitatory neurons reveals cell-type-specific chromatin loops and TADs, overlapping with differentially expressed genes. ArcFISH is freely available at https://hyuyu104.github.io/ArcFISH.

genomics↗

bioRxiv2025

A Reference Atlas of the Human Dorsal Root Ganglion

Somatosensory perception largely emerges from diverse peripheral sensory neurons whose cell bodies reside in dorsal root ganglia (DRG). Damage or dysfunction of DRG neurons is a major cause of chronic pain and sensory loss. In mice, deep single-cell transcriptomic profiling and genetically defined models have offered important clues into DRG function, but in humans, the cellular and molecular landscape of DRG neurons remains less understood. Here, we constructed a reference cell atlas of the human DRG by profiling transcriptomes of cells and nuclei from 126 donors sampled across cervical, thoracic, and lumbar DRGs. This atlas resolves 22 neuronal subtypes, including known and previously unrecognized subtypes linked to nociception, mechanosensation, thermosensation, and proprioception, as well as 10 types of non-neuronal cells. Cross-species integration, spatial transcriptomics, and microneurography enabled cell-type-specific comparisons of soma size and conduction velocity between species. Human DRG somata are larger across all cell types than their mouse counterparts, and the conduction velocities of human hair follicle innervating A-fibers are faster than in mice, suggesting a functional shift in rapid mechanical detection in humans. This integrated human DRG reference cell atlas provides a resource for exploring new molecular and physiological features of human DRG, which could help identify new strategies for treating chronic pain and other diseases of the peripheral nervous system.

neuroscience↗

bioRxiv2025

Evaluating Field Trial Designs for Genetically Modified Mosquito Interventions: An In-Silico Simulation Approach

Mosquito control strategies based on the mass release of modified males, such as genetically modified mosquitoes (GMMs), aim to suppress wild populations by impairing reproduction. Evaluating these interventions requires resource-intensive field trials, but a lack of standardized implementation practices, particularly regarding release ratios of modified males to wild female mosquitoes and trial timing, has led to variable outcomes. This studys objective is to propose a modeling tool for the "in-silico" simulation of trial designs before field implementation. To this aim, we developed an agent-based model of mosquito population dynamics. As a case study, we calibrated the model using 2019-2023 Aedes aegypti surveillance data from Miami-Dade County, Florida, and compared two GMM trial designs as illustrative examples. Our results show that depending on the implementation choices (e.g., trial start date and duration, release ratio), trials yield highly variable outcomes. For example, changing the start date while fixing all other implementation details can lead to effectiveness between 50% and 90%. Our findings suggest that "in-silico" simulation is a valuable tool for improving trial protocol design, allowing stakeholders to test strategies and reduce outcome uncertainty before committing to a fieldwork experiment.

ecology↗

bioRxiv2025

End-to-end genomic prediction: direct prediction of images and text from genome-wide molecular markers

BackgroundMethods to predict the heritable component of phenotypes from genetic markers, collectively known as genomic prediction, have been widely applied in plant and animal breeding and in human genetics. Currently, genomic prediction is limited to numeric phenotypes. In some cases, though, plant and animal phenotypes are better understood through images and text rather than numbers. The current best practice for incorporating images and text in genomic prediction is to first extract scalar numeric phenotypes from images and text, and then to perform genomic prediction on the numeric phenotypes. While this approach is effective for some traits, it involves discarding most of the information in the image or text, including potentially useful information. Additionally, numeric phenotypes derived from images and text may not be as interpretable as the images and text themselves. ApproachWe present a novel approach for predicting images and text from SNP markers, which we refer to as end-to-end genomic prediction, and validate this approach using genotypes, text, and image phenotypes derived from a strawberry (Fragaria x ananassa) diversity panel. Our approach combines nonlinear latent space encoding with linear genomic prediction to generate accurate breeding values for a high-dimensional phenotype, for example, images or text. ResultsFor both genome-to-image and genome-to-text prediction, we found that predicting images and text and then extracting numeric traits from them was in some cases as accurate as directly predicting extracted numeric phenotypes, demonstrating for the first time that genome-to-image prediction accuracy can be comparable to conventional genomic prediction accuracy. ConclusionsBased on our proof-of-concept using the same core end-to-end method in both images and text, we believe end-to-end genomic prediction could be of use in a wide range of visual and multidimensional phenotypes in plants and animals, although further work to improve the accuracy of the embedding and genomic prediction steps is needed.

genomics↗