Plant methods

Rapid and objective assessment of foliar disease is important for screening host responses and evaluating novel control agents. We developed a complementary suite of low-cost methods for quantitative assessment of wheat leaf rust, combining an ImageJ-based disease quantification workflow (RONUM) with detached-leaf and soil-free whole-plant infection assays. RONUM enabled semi-automated quantification of disease metrics following user-guided calibration. Against manually corrected reference image sets from whole plant leaf scans and detached leaves, disease estimates showed strong agreement with reference measurements and were reproducible following independent calibration by multiple users, and by pixel agreement optimisation to the reference masks. Benchmarked against another published ImageJ workflow, RONUM also showed closer agreement with manually corrected measurements, particularly for images containing chlorosis, necrosis or poor leaf health. Post-processing options were quantitatively assessed, showing little influence over total pustule-area estimates but substantial effects upon pustule count estimates. Two complementary infection assays were presented to demonstrate example applications of the quantification workflow: a soil-free whole-plant assay for non-destructive tracking of leaf rust development over an 18-day infection time course; and a benchtop-scale detached-leaf assay. The detached leaf assay provided reproducible disease quantification across 13 wheat lines with contrasting disease phenotypes. Exploratory quantification of chlorosis: pustule area ratios substantially separated the susceptible lines from those harbouring pathotype-specific Lr resistance, showcasing a potential application for collecting biologically relevant phenotype data typically unavailable to image-based quantitation. Finally, a modified detached leaf assay with nylon-mesh treatment reservoir was developed and assessed by establishing a dose-response curve to a propiconazole fungicide formulation, with an estimated IC₅₀ of 20.49 ppm formulation (95

Occlusion is a major factor limiting accurate three-dimensional (3D) wheat phenotyping. In natural growth conditions, overlapping spikes, leaves, and stems often make only partial target regions visible in single-view images, hindering complete and reliable 3D reconstruction. To address this problem, this study proposes an amodal completion-assisted, sequential framework for single-view 3D reconstruction of occluded wheat. The framework first uses visible prompts to recover the complete appearance and structural cues of occluded targets, and then feeds the completed images into single-view 3D reconstruction models to generate complete 3D structures. We construct the MMWO (Multi-view Multi-instance Wheat Occlusion) dataset from MMW, which is captured under controlled indoor scenarios, by synthesizing diverse occlusion samples through organ-level cutouts, random geometric transformations, and region-constrained pasting, with annotations including visible masks, occlusion masks, and complete target images. Six representative reconstruction methods, including Direct3D, Real3D, SF3D, Spar3D, TRELLIS.2, and Hunyuan3D, are systematically evaluated. Hunyuan3D achieves the best geometric performance, with the lowest mean CD- L_1 and CD- L_2 values of 0.1286 and 0.0536, and the highest mean F-score of 0.5668. SF3D achieves the best rendering quality in terms of PSNR, SSIM, and LPIPS. In addition, Pix2Gestalt completion reduces the estimation errors of spike length, width, and area from 9.31

Pea seed-borne mosaic virus (PSbMV) is an important seed-transmitted pathogen of pulse crops that can reduce yield and seed quality while remaining difficult to detect using visual inspection. Conventional diagnostic methods are accurate, but destructive, labor-intensive, and unsuitable for large-scale seed screening. In this study, we evaluated hyperspectral imaging (HSI) as a non-destructive approach for detecting PSbMV infection in faba bean (Vicia faba) seeds. Individual seeds were imaged using visible-near-infrared (Vis–NIR) and shortwave-infrared (SWIR) hyperspectral systems (400–1700 nm), and directional reflectance spectra were extracted after reflectance calibration and preprocessing. Principal Component Analysis (PCA) revealed subtle but consistent spectral differences between infected and healthy seeds. Among the evaluated models and preprocessing methods, a support vector machine combined with standard normal variate preprocessing achieved the best performance, with a mean fivefold cross-validation accuracy of 97.2

Abstract Background Understanding the structure of plant seeds cultivated for human consumption and food manufacturing is vital to provide sustainable products as well as to investigate early growth stages. This includes structural variation between different plant species, varieties and cultivars depending on genetic setup, as well as structural modifications upon germination, aging and storing or seed treatment during processing. For plant seeds as multi-component biological materials, structural characterization must extend across multiple length scales, from molecular organization to cellular architecture. Results We apply scanning Small- and Wide-Angle X-ray Scattering (SWAXS) and X-ray Fluorescence (XRF) on yellow pea seeds to combine local structural information on the molecular scale with imaging of cellular structures on the micrometer scale, enabling a comprehensive analysis of hierarchical organization. To identify and characterize heterogeneous regions within the pea seeds, we implement a fitting-free, data-driven segmentation and analysis workflow based on machine learning tools. This approach allows for classification of structurally distinct domains and enables quantitative comparison across samples without relying on predefined models. Furthermore, we incorporate multi-modal analysis by combining structural imaging with complementary elemental information obtained from XRF. The integration of compositional and structural data provides deeper insight into structure-composition relationships. Conclusions This multi-scale, multi-modal approach opens new possibilities for investigating hierarchical structures and their development under diverse conditions and enables systematic comparison between different species or seeds at different developmental stages or exposed to different processing steps. The approach is broadly applicable to various kinds of samples and other hierarchically organized biological materials, which makes it a valuable technique for plant science as well as plant-based food science.

The present study provides the first comprehensive protocol for the in vitro propagation of Opposite-leaved Pondweed (Groenlandia densa (L.) Fourr.), a critically endangered aquatic macrophyte in Poland that requires active conservation measures. Submerged shoots of G. densa were surface-sterilised by immersion in 70

Nodal roots are central to water and nutrient uptake, anchorage, and resistance to lodging in grasses. These functions are especially important under variable and severe weather conditions that impact plant performance and crop stability. However, controlled induction and quantitative analysis of nodal roots under controlled conditions remain a major technical challenge. To address this limitation, we developed a hydroponic growth and analytical platform that enables reproducible induction and quantitative analysis of shoot-borne nodal roots beyond early seedling stages. The hydroponic system supported stable nodal root development and enabled controlled chemical perturbations. When combined with histological and quantitative imaging approaches, the platform resolved tissue-specific cell wall properties across defined cell types. Using Brachypodium distachyon and Triticum aestivum (wheat), we found that phytohormone perturbations produced distinct effects on leaf nodal root development and anatomy. Treatment with GA3 or trans-zeatin increased secondary wall thickening in both cortical and vascular tissues. Perturbation of auxin application primarily affected root initiation and elongation with limited effects on wall deposition. Jasmonate signaling altered cell wall properties primarily in outer cortical tissues, revealing an uncoupling of lignin accumulation and wall thickness, while vascular wall thickness remained largely unchanged. The platform was successfully applied to wheat, where GA3 treatment induced cortical secondary wall thickening similar to that observed in B. distachyon. This platform provides a scalable and accessible method for studying nodal root development and secondary cell wall deposition in grasses. The results highlight strong cell type-specific regulation of wall thickening and demonstrate the utility of the platform for comparative studies of root anatomy in grasses. By linking controlled perturbations with quantitative anatomical measurements, this approach enables comparative analysis of root traits relevant to plant stability and performance and provides a foundation for future studies of root development in grasses.

In Italy, grapevine production is central to agricultural productivity and cultural heritage, but it is increasingly threatened by Xylella fastidiosa subsp. fastidiosa (Xff) the causal agent of Pierce’s disease under field conditions infection. The recently detected Xff ST1 in Apulia colonizes grapevine xylem vessels, impairing water and nutrient transport and leading to leaf scorch, wilting, and vine mortality. In this study, following inoculation with Xff, a rapid screening protocol based on shoot culture and micropropagation was developed as a preliminary in vitro approach to evaluate disease progression, symptom development under in vitro conditions, in eight Calabrian grapevine varieties: ‘Aglianico’, ‘Bianca Antica’, ‘Gaglioppo’, ‘Greco Bianco’, ‘Lacrima Nera’, ‘Magliocco Canino’, ‘Magliocco Dolce’, and ‘Zagarese’. The results highlighted significant differences in estimated bacterial population levels, expressed as log₁₀ CFU equivalents mL⁻¹, among varieties, with higher colonization observed in ‘Aglianico’, ‘Bianca Antica’, and ‘Magliocco Dolce’, ranging from approximately 4.0 to 4.5 Log₁₀ CFU mL⁻¹. Differential susceptibility to Xff was also evident over the 60-day post-inoculation period. Notably, ‘Aglianico’ and ‘Magliocco Dolce’ were the most symptomatic varieties, whereas ‘Greco Bianco’ and ‘Magliocco Canino’ showed the lowest symptom development, displaying the lowest rAUDPC values, 0.01 and 0.03, respectively, suggesting lower susceptibility under the experimental conditions tested. Substantial differences in symptom progression were also observed, with distinct temporal patterns of disease development among the eight varieties over the 60-day period. This study provides a comparative regional proof-of-concept for assessing baseline susceptibility variations among previously uncharacterized Calabrian grapevine genotypes. Furthermore, it presents a rapid in vitro method for evaluating Vitis vinifera’s susceptibility to Xff, potentially serving as a significant preliminary screening tool to support subsequent validation studies and field assessments.

Sweet potato foot rot is difficult to detect at the pre-symptomatic stage because visible symptoms are minimal or absent, whereas conventional approaches are often destructive or unsuitable for rapid screening. In this study, a UV fluorescence imaging workflow was developed for pre-symptomatic evaluation of sweet potato foot rot based on host-response-associated autofluorescence. The method uses two-band fluorescence acquisition at 460 and 520 nm under 375 nm excitation, followed by pixel-wise and distribution-based image analysis. Rather than relying only on localized bright regions or overall signal intensity, the workflow represents tissue states using multi-channel pixel distributions and exploratory quantitative descriptors. Scatter distributions were used to visualize joint channel behavior, and exploratory single- and dual-band descriptors were used for quantitative comparison between Control and Stage 1 samples. Pre-symptomatic Stage 1 samples, which were macroscopically indistinguishable from Controls, exhibited characteristic distribution features that differed from Control patterns under the same acquisition and preprocessing conditions. Scopoletin injection reproduced Stage 1-like fluorescence distribution patterns, supporting consistency with phytoalexin-associated autofluorescence. The workflow was further applicable to a second sweet potato cultivar without cultivar-specific parameter tuning, and complementary ATR-FTIR measurements supported chemical interpretation. These results establish a distribution-based fluorescence imaging framework for pre-symptomatic sweet potato foot rot evaluation and provide a basis for future automated analysis within sampling-based screening workflows.

ATP-binding cassette (ABC) transporters form one of the largest and most functionally diverse families of membrane proteins in plants, with members of the ABCG subfamily playing central roles in the transport of specialized metabolites and stress adaptation. In legumes functional analyses of these transporters have been limited by functional redundancy, accessibility of T-DNA inserted mutant lines, and the constraints of RNA interference–based approaches. To address these limitations, we used Medicago truncatula ABCG46 transporter and established hairy root–based CRISPR/Cas9 genome-editing platform that enables rapid empirical validation of single guide RNAs prior to stable transformation. Two guide RNAs targeting distinct exons of MtABCG46 were evaluated in 70 independent hairy root lines, revealing strong editing activity for one guide, while the second proved non-functional despite favourable in silico predictions. Sequence analysis identified a range of insertions and deletions, including a homozygous biallelic mutant carrying frame-shifting deletions predicted to abolish transporter function. Using the validated guide RNA, stable Agrobacterium tumefaciens–mediated transformation yielded heritable mutations, and transgene-free knockout lines were recovered in subsequent generations. This study establishes a robust hairy root-based gRNA validation platform for M. truncatula, generates stable and heritable mtabcg46 knockout lines, and provides a scalable framework for functional characterization of ABCG transporters in specialized metabolism and plant defense.

In plant synthetic biology, the emergence of Golden Gate (GG) cloning has led to the rapid development of numerous cloning kits that enable the standardized assembly of genetic modules and transcriptional units through a single, one-pot reaction. However, this rapid and somewhat uncoordinated expansion has introduced challenges, particularly regarding compatibility between cloning kits, which can hinder adoption and routine use in new laboratories. To address these recurring issues, this review provides a comprehensive overview of more than 25 well-established plant-specific GG cloning kits, along with an in-depth examination of the compatibility of their available modules, with the goal of improving interoperability across systems. To support a more integrated and forward-looking development of future GG-based efforts, we present key approaches for increasing system harmonization, including several domestication strategies, methylase-assisted hierarchical DNA assembly, and the inclusion of multiple cloning sites. As a whole, this review aims to streamline cloning workflows and reduce technical barriers for new plant synthetic biologists in pursuit of increasing the throughput of their experiments.

Allele-specific expression analysis can reveal cis-regulatory differences (e.g., promoter variants, epigenetic changes) that cause imbalanced gene expression between haplotypes. Haplotype-resolved reference genomes and long-read RNA sequencing enable allele-specific expression analysis at gene and isoform-levels. However, existing tools are largely restricted to short-read RNA sequencing data and diploid organisms. We developed LongPolyASE, an end-to-end computational framework for allele-specific gene and isoform expression analysis in diploid and polyploid organisms using long-read RNA sequencing, consisting of three components: Syntelogfinder, for identifying syntenic gene relationships and annotation inconsistencies; longrnaseq, for novel isoform discovery and haplotype-level quantification; and PolyASE, for statistical testing and visualization of allelic imbalance and isoform usage. We applied LongPolyASE to diploid rice, autotetraploid potato, allotetraploid rapeseed, and allooctoploid strawberry using Oxford Nanopore and PacBio long-read RNA-seq. The framework enabled identification of cis-regulatory variation, tissue-specific trans-regulatory effects, differential isoform usage, and haplotype-specific splicing differences. In addition, it facilitated the discovery of novel transcripts and genes with potential functional relevance in plant development. LongPolyASE addresses a key methodological gap by enabling allele-specific expression analysis in polyploid organisms using long-read RNA sequencing. By combining haplotype-aware quantification with isoform-level resolution in a reproducible workflow, the framework provides a practical tool for plant researchers working with complex genomes. Its application to crop species highlights its potential to support the identification of regulatory variation and candidate targets for plant breeding.

Salinity is a major abiotic stress that negatively affects nearly all plant species at all stages of growth. Drought and poor-quality irrigation cause high soil salinity and salt accumulation via evaporation, reducing crop productivity. Despite its critical importance, the spatial localization of salt ions and associated biochemical changes within plants experiencing high salinity remains largely unknown. In this study, we developed a multimodal imaging pipeline to understand the impact of salinity on the pistachio rootstock UCB-1 (Pistacia atlantica x Pistacia integerrima). We directly link biochemical fingerprints in stem tissue architecture with salt ion localization to provide insights into the strategies pistachio uses to tolerate salinity. We observed that Pistacia spp. exposed to high salt conditions accumulated Ca, Si, Cl, Al and Mg as hotspots within the pith, compared to the control (of which only Ca and Al co-locate). In contrast, there was a decrease in K between the control and salinity treatment. Hotspots of amide I and II were present in the cortex and pith of the salinity treated sample. Additionally, the salinity treatment resulted in an increased abundance of pectin and carbohydrates within the pith compared to the control, and the abundance of esters/carboxylic acid was greater in the salinity treatment. We determined that Cl and K, S and P, and biochemical components polysaccharide and pectin, esters and carboxylic acid, amide I and cellulose are the strongest drivers of salinity-treatment induced variability. In the cortex and phloem/xylem, a negative K-Ca correlation decreases in the salinity treatment. Several hotspots of elements and amide I (proteins) appear under salinity treatment, particularly in the cortex, suggesting an increase in the production of stress-related proteins (in response to high Cl) and/or structural proteins (i.e. Ca). Together, these results indicate that pistachio responds to salinity through ion compartmentalization coupled with a targeted biochemical adjustment, rather than a broadscale tissue-wide response. Overall, these novel, spatially resolved pixel-registered multimodal imaging data provide an enabling platform to understand the mechanisms of salinity tolerance in Pistacia spp and can be broadly applied to studying stress-related phenotype response in various plant tissues.

Post-curing nitrogen content (cNg) and nicotine content (cNt) constitute fundamental chemical attributes that govern the quality of flue-cured tobacco leaves. In this study, we investigate the use of unmanned aerial vehicle (UAV)-borne hyperspectral imagery in conjunction with advanced sequence-learning models to obtain canopy-level spectral information and to quantitatively predict cNg and cNt at the field scale. Nevertheless, because field experiments conducted in different growing seasons follow heterogeneous observation schedules, the resulting time series differ in both length and temporal sampling density, which poses a substantial obstacle to the direct deployment of recurrent neural-network models. To mitigate this limitation, we devised three autoencoder-based temporal-alignment schemes to reconcile the temporal dimension of multi-year spectral–phenotypic sequences, namely a fully connected autoencoder (AEF), a one-dimensional convolutional autoencoder (AEC), and a long short-term memory (LSTM) autoencoder (AEL). Within this framework, the autoencoders project the original multi-temporal observations into compact latent representations and subsequently reconstruct them on a standardized temporal grid, thereby preserving salient spectral–phenotypic information while enforcing dimensional consistency across years. For comparative analysis, we additionally considered a straightforward time-step removal (TSR) procedure, which discards non-overlapping observation dates among different years and thus serves as a baseline temporal-harmonization strategy. On the temporally aligned sequences, two representative recurrent-neural architectures—long short-term memory (LSTM) networks and gated recurrent unit (GRU) networks—were trained to establish predictive models for cNg and cNt. Overall, models trained on inputs aligned by the autoencoder-based schemes exhibited markedly higher predictive skill than their TSR-based counterparts, with the convolutional and LSTM autoencoders consistently yielding greater improvements than the fully connected variant. Among all evaluated configurations, the combinations of AEL with LSTM and with GRU delivered the best performance, attaining coefficients of determination of 0.73 for cNg and 0.56 for cNt on the independent test set. Taken together, these results indicate that the proposed autoencoder-based temporal-alignment framework can effectively distil informative features from multi-year UAV hyperspectral and phenotypic observations and constitutes a promising tool for the quantitative prediction of post-curing quality indicators in flue-cured tobacco production.

BeanGPT is a domain-specific retrieval augmented generation system designed to support research and breeding decisions in common bean (Phaseolus vulgaris L.) by transforming natural language questions into citation-backed, verifiable answers. The platform integrates a large, curated corpus of legume-focused peer-reviewed literature with structured multi-year agronomic trial records collected across diverse environments, climate projections extending to 2090 under multiple emission scenarios, and standardized cultivar nomenclature to resolve naming inconsistencies across datasets and publications. BeanGPT combines semantic retrieval from a vector database with intent-based query routing and structured parameter extraction to direct questions to genetics, field performance analytics, or climate modules. To reduce errors that commonly occur in general-purpose language models, BeanGPT incorporates a genomic index that enables constant time membership lookup of gene and protein identifiers against authoritative resources, ensuring that molecular entities are either validated or clearly flagged as literature-derived. The system is implemented with a streaming web interface and an asynchronous backend that supports concurrent users and can generate interactive visualizations through automated Plotly code generation. Beta testing demonstrated strong retrieval relevance, low response latency, reliable gene verification, and high citation precision, indicating that domain-grounded RAG can improve accuracy and usability for Phaseolus vulgaris research.

Eggplant (Solanum melongena) is a major cash crop, yet field detection of its pests and diseases remains difficult because disease evidence is simultaneously occluded by foliage, blurred at lesion boundaries, and highly variable in scale. Fruit rot is especially challenging: the waxy epidermis and purple anthocyanin-rich surface reduce chromatic contrast, while infection spreads gradually from water-soaked tissue to necrotic tissue, producing diffuse borders between diseased and healthy regions. In this work, we reframe eggplant disease detection through a context-boundary-scale coupling principle, which states that accurate field detection should jointly model incomplete contextual cues, ambiguous lesion boundaries, and scale-varying symptom morphology rather than optimize these cues independently. ISA-YOLO is proposed as an implementation of this principle on top of YOLOv13 through coordinated context modeling, boundary-aware aggregation, and progressive multi-scale fusion. Experiments on two public datasets show that ISA-YOLO achieves 78.1 and 77.7

Precise characterization of gene expression patterns across temporal, cellular, and tissue-specific contexts is fundamental to understanding plant development and function. Recent advances in ClearSee-based tissue clearing have enabled high-resolution visualization of internal structures and fluorescent reporter signals in plant tissues. Although hand sectioning can provide optical access to tissues that are not amenable to whole-mount clearing, its application to submillimeter-scale and fragile Arabidopsis organs and tissues, including developing inflorescence apices, flowers, fruits, and organ boundaries, has remained limited. Consequently, analysis of these tissues has largely depended on specialized microdissection techniques and labor-intensive histological workflows, such as wax- or resin-embedded microtomy, which restrict throughput, accessibility, and routine use. We developed and optimized a simple hand-sectioning and imaging method that enables routine visualization of anatomical organization and gene expression patterns at cellular resolution in small, fragile Arabidopsis tissues. This method relies only on gentle manual tissue processing under a stereomicroscope and readily available reagents, allowing reproducible preparation of delicate tissues without the need for embedding or specialized equipment. Combined with ClearSee-based clearing and fluorescent reporters, the approach enables high-resolution imaging of internal tissue architecture and gene expression, while preserving tissue integrity and fluorescence signals that are often compromised during conventional embedding and microtomy procedures. Our method substantially reduces technical complexity, costs, preparation time, and labor associated with cellular-resolution imaging of small, fragile plant tissues. By providing a simple, scalable, and accessible alternative to conventional histological workflows, this approach facilitates routine analysis of internal developmental processes across diverse plant species.

Functional genomics in plant biology relies on the generation, reuse, and long-term management of large numbers of plasmids produced through diverse cloning strategies. As collections expand across users and projects, laboratories face increasing challenges in organization, traceability, and preservation of construction histories. Existing cloning and sequence-design software supports plasmid design but does not address collaborative, laboratory-wide plasmid management. We developed PlasmiDB, an open-source, web-based database for managing plasmid collections in multi-user research environments. PlasmiDB provides structured storage of plasmid metadata, explicit tracking of plasmid genealogy, and traceability throughout the plasmid life cycle, including inter-laboratory exchanges. The system implements project-based access control and supports collaborative workflows involving staff, students, and core facilities. Implemented using a standard LAMP architecture and deployable via Docker, PlasmiDB is designed for extensibility without modification of the core database schema. A gene module links genetic targets to associated plasmids, primers, and CRISPR reagents, improving coherence between molecular constructs and experimental objectives. In our laboratory, PlasmiDB currently manages nearly 700 plasmids and facilitates reporting through automated declaration file generation. PlasmiDB complements existing cloning tools by providing traceability, collaboration support, and long-term data stewardship for plant molecular biology laboratories. The source code and Docker image are publicly available.

Quantification of rice root anatomical traits such as cortical aerenchyma lacunae is key to understanding rice adaptation to diverse water regimes and to support climate-smart breeding. Aerenchyma lacunae contributes to rice internal gas transport and influences methane emissions from flooded systems and can also limit rice water conductivity. It could be an interesting anatomical trait for breeding, however, large-scale anatomical phenotyping remains limited because manual analysis of root cross-sections is labor-intensive, subjective, and difficult to scale across heterogeneous imaging conditions. Existing pipelines often require parameter tuning and do not generalize well across environments. We developed a deep learning pipeline based on a vision transformer architecture to automatically segment rice root cross-sections and quantify cortical aerenchyma lacunae. The model was trained on 1,760 annotated images collected across multiple countries, growth stages, cultivation systems, and experimental contexts, using a collaboratively defined annotation protocol. The final model achieved high segmentation accuracy, with mean intersection over union values exceeding 0.92 for cortical tissues and lacunae. Quantification of the lacuna-to-cortex ratio showed strong agreement with manual annotations, with a coefficient of determination of 0.98 on an independent test set. An independent expert review indicated that model predictions were at least as consistent as manual annotations and reduced large annotation inconsistencies. The pipeline is released as open-source software and includes an interactive online demonstrator, and is accompanied by an online test dataset to support testing and reproducibility. Application across six experimental use cases revealed reproducible differences in aerenchyma lacunae across genotypes, water regimes, environments, and developmental stages. This work provides a robust, scalable, and transferable tool for automated root anatomical phenotyping under heterogeneous experimental conditions. Transformer-based segmentation enables consistent and high-throughput quantification of lacunae, facilitating integration of these anatomical traits into breeding, physiological studies, and climate-smart crop improvement programs.

Tung tree (Vernicia fordii) serves as an economically important woody oilseed plant. High-quality protoplasts are essential for applications such as somatic hybridization, studying cell signal transduction and rapid gene function analysis. A reliable protocol for protoplast isolation and transient gene expression was developed using petals, anther-derived callus, and root tips as source materials. Through systematic optimization of key parameters such as enzyme combinations, osmotic pressure, digestion duration, and centrifugation speed, we successfully established tissue-specific isolation protocols. To ensure the availability of widely applicable experimental materials, protoplasts isolated from anther-derived callus were used to optimize transient transformation conditions. Under the optimized parameters (20 µg plasmid, 40

Genomic prediction (GP) plays a pivotal role in expediting genetic gains for crop breeding. Recently, deep learning models has attracted growing attention in this field due to its superior performance over traditional models. However, most advanced models rely on simplified numerical representations of genomic variants and treat individuals in isolation, failing to capture the full complexity of genetic variation and the genetic relatedness among individuals. In response, we propose RAGP, a deep learning model driven by a retrieval-augmented mechanism composed of two synergistic components: (1) an embedding-based retrieval module that identifies a group of nearest neighbors highly relevant to the target sample as references, employs a phenotype-aware encoder to refine the representation space and capture informative genetic structures; (2) an augmentation module that integrates the retrieved references to enhance the target representation. The enriched representation is then propagated through a regression network for final prediction. This retrieval-guided strategy enables more informed and fine-grained utilization of genomic data and strengthens the model’s capacity to capture individual-level genetic relationships. Extensive experiments reveal that RAGP achieves consistently superior predictive performance compared to both conventional methods and state-of-the-art deep learning models. By incorporating retrieval-augmented mechanisms, RAGP achieves substantial improvements in predictive accuracy and offers a promising direction for advancing genomic selection in crop breeding. Our code is available on https://github.com/l00907l/RAGP.