Visual analysis method and system for rice multi-tissue single cell expression profile

Abstract

The invention relates to the technical field of bioinformatics, and provides a visual analysis method and system for a rice multi-tissue single cell expression profile. The method comprises the following steps: comparing sequencing data of an original single cell transcriptome of a rice tissue to obtain a standardized transcriptome data set; performing batch effect correction and integration on the standardized transcriptome data set to obtain a whole plant expression matrix; performing cell type annotation on the whole plant expression matrix to obtain a cell type annotation system; carrying out visual dimension reduction processing on the whole plant expression matrix fused with the cell type annotation system, and carrying out co-expression network construction to obtain a modular tissue correlation analysis model; and establishing an interaction end based on the module organization correlation analysis model, and realizing data visualization analysis through the interaction end. The invention provides a one-stop analysis platform for rice cell heterogeneity research, functional gene mining and molecular breeding.

Description

Technical Field

[0001] The present invention relates to the technical field of bioinformatics, and in particular to a method and system for visualizing and analyzing rice multi-tissue single-cell expression profiles. Background Art

[0002] Single-cell RNA sequencing (scRNA-seq) technology, a key tool in modern biological research, has been widely used in analyzing plant cell heterogeneity and developmental biology. In rice functional genomics, single-cell transcriptomics technology can analyze gene expression patterns across different tissues and developmental stages at single-cell resolution, providing unprecedented opportunities for understanding rice growth and development mechanisms, cell differentiation processes, and gene regulatory networks.

[0003] However, in rice research, the application of single-cell RNA sequencing (scRNA-seq) technology faces bottlenecks: existing rice single-cell data are scattered across different studies and lack standardized integration across developmental stages and multiple tissues; mainstream tools (such as SCope) only support single-dataset presentation, cannot achieve cross-tissue gene expression comparison, and lack crop-specific analysis modules; cell type identification relies on manual marker gene screening, which is time-consuming and highly subjective, and there is no standardized process for root cell subtype annotation in particular. In addition, while current platforms such as the Plant Single Cell Atlas provide some crop data, they have significant flaws: the interactive interface only supports static charts and cannot retrieve gene expression patterns in real time; and advanced analysis functions such as co-expression networks are not integrated. These shortcomings seriously hinder the analysis of rice cell function and the precise design of breeding processes. Summary of the Invention

[0004] The present invention provides a method and system for visual analysis of rice multi-tissue single-cell expression profiles, to address the deficiencies of the prior art.

[0005] The present invention provides a method for visualizing and analyzing rice multi-tissue single-cell expression profiles, comprising:

[0006] S1: Align the sequencing data of the original single-cell transcriptome of rice tissue to obtain a standardized transcriptome dataset;

[0007] S2: performing batch effect correction on the normalized transcriptome dataset and integrating the dataset to obtain a whole-plant expression matrix;

[0008] S3: performing cell type annotation on the whole plant expression matrix to obtain a cell type annotation system;

[0009] S4: Visualize and reduce the dimensionality of the whole-plant expression matrix integrated with the cell type annotation system, and construct a co-expression network to obtain a modular tissue correlation analysis model;

[0010] S5: An interactive terminal is established based on the module organization correlation analysis model, and data visualization analysis is achieved through the interactive terminal.

[0011] According to a visualization analysis method for rice multi-tissue single-cell expression profiles provided by the present invention, in step S1, the rice tissues include: rice crown roots, rice stems, rice leaves, rice flag leaves, rice stem tips, rice tiller buds, rice panicles, and rice seeds.

[0012] According to the method for visualizing and analyzing rice multi-tissue single-cell expression profiles provided by the present invention, step S2 further comprises:

[0013] S21: performing normalization processing on the gene expression data in the standardized transcriptome dataset to obtain normalized expression data;

[0014] S22: Perform multi-tissue integration on the normalized expression data based on the Harmony algorithm to establish a whole-plant expression matrix.

[0015] According to the method for visualizing and analyzing rice multi-tissue single-cell expression profiles provided by the present invention, step S3 further comprises:

[0016] S31: Construct a marker gene database based on the collected rice marker genes and homologous marker genes;

[0017] S32: Based on the marker gene database, the cells in the whole plant expression matrix are annotated by type through in sit experimental verification to obtain a cell type annotation system.

[0018] According to the method for visualizing and analyzing rice multi-tissue single-cell expression profiles provided by the present invention, step S4 further comprises:

[0019] S41: mapping the whole plant expression matrix to a two-dimensional space to obtain a two-dimensional space representation;

[0020] S42: fusing the two-dimensional spatial representation with the standardized cell type annotation system to generate a cell cluster distribution map;

[0021] S43: calculating the correlation between genes in the whole plant expression matrix to obtain a gene correlation matrix;

[0022] S44: Based on the cell cluster distribution map and the gene correlation matrix, a module tissue correlation analysis model is constructed.

[0023] A method for visualizing and analyzing rice multi-tissue single-cell expression profiles provided by the present invention further comprises:

[0024] Step S6: Based on the interactive end, the root cell type is predicted for the input expression matrix through the pre-trained XGBoost model to obtain the root cell probability prediction result.

[0025] According to the method for visualizing and analyzing rice multi-tissue single-cell expression profiles provided by the present invention, step S6 further comprises:

[0026] S61: receiving an input expression matrix;

[0027] S62: preprocessing the input expression matrix using a scanpy normalization algorithm to obtain preprocessed expression data;

[0028] S63: Using the pre-trained XGBoost model, cell type classification is performed based on the highly variable gene features in the pre-processed expression data to obtain a root cell probability prediction result.

[0029] The present invention also provides a visualization analysis system for rice multi-tissue single-cell expression profiles, comprising:

[0030] The front-end interactive interface is a responsive front-end built on the React framework, including a main menu, an organizational map portal, and a gene search panel with a three-tiered navigation architecture;

[0031] Backend service architecture, which is a static resource server deployed through Apache / 2.4.63 to support high-throughput access and data transmission;

[0032] A tissue map browsing module, comprising an input layer, a visualization engine, and an output layer, for displaying and processing multi-dimensional tissue cell clustering maps based on the UMAP dimensionality reduction algorithm;

[0033] A gene retrieval module, comprising an input layer, an analysis engine, and an output layer, for retrieving and analyzing the cross-tissue expression patterns of target genes based on MSU / RAP gene IDs;

[0034] Network analysis module: used to construct single-cell co-expression networks based on the hdWGCNA algorithm and generate module tissue correlation analysis models.

[0035] A visualization analysis system for rice multi-tissue single-cell expression profiles provided by the present invention further includes:

[0036] The RSCA annotation tool module is built based on a pre-trained XGBoost model and is used to automatically annotate root cell types on 10x Genomics expression matrices uploaded by users.

[0037] According to the present invention, a visualization analysis system for rice multi-tissue single-cell expression profiles is provided. The input layer of the tissue map browsing module supports user selection modes including specific tissue and whole plant views. The visualization engine of the tissue map browsing module supports scaled UMAP dimensionality reduction maps based on SVG vector graphics to display cell cluster distribution. The output layer of the tissue map browsing module provides export and metadata download functions.

[0038] The input layer of the gene retrieval module supports gene ID input and tissue screening. The analysis engine of the gene retrieval module quantifies the TPM expression intensity of the target gene in the selected cell type through violin plots and generates a cross-tissue expression pattern heat map. The output layer of the gene retrieval module displays the single-cell expression data through interactive charts.

[0039] The network analysis module constructs a single-cell co-expression network based on preset soft threshold parameters, displays the core regulatory modules through a regulatory network diagram, annotates related genes based on module organization association analysis, and supports the export of network format files.

[0040] The present invention provides a method and system for visual analysis of rice multi-tissue single-cell expression profiles. CellRanger is used to compare and process raw scRNA-seq data from rice tissues. Combined with the multi-tissue data integration processing using the Seurat normalization algorithm and the Harmony algorithm, this method can effectively eliminate batch effects and technical noise between different tissues, ensuring data consistency and comparability. This method constructs a whole-plant expression matrix containing 116,564 high-quality cells, providing a high-quality data foundation for subsequent analysis. Furthermore, a standardized cell type annotation system based on rice marker genes and homologous marker genes significantly improves the accuracy and consistency of cell type identification, avoids the subjectivity and inconsistency problems in traditional manual annotation methods, and ensures good comparability of cell type annotation results between different studies. Furthermore, through visual dimensionality reduction processing using the UMAP dimensionality reduction algorithm, high-dimensional single-cell expression data can be mapped to a two-dimensional space, generating an intuitive and clear cell cluster distribution map. This allows researchers to quickly identify the distribution patterns and relationships between different cell types, greatly improving the efficiency and accuracy of data interpretation. The hdWGCNA algorithm, based on soft-threshold parameter co-expression network construction and processing, can identify biologically significant gene co-expression modules, revealing regulatory relationships and functional connections between genes, providing an important basis for understanding rice gene regulatory networks and cellular functions. Furthermore, the root cell type prediction processing of the pre-trained XGBoost model achieves high-precision automatic annotation, greatly reducing the workload and time cost of cell type identification and improving the objectivity and repeatability of the annotation results. Furthermore, the system provided by the present invention is based on a B / S architecture and a responsive front-end design, supporting multi-user concurrent access and real-time interactive analysis, providing a convenient and efficient data analysis platform for rice researchers worldwide, promoting collaboration and development in rice functional genomics research, and accelerating the progress of rice molecular breeding and variety improvement. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the technical solutions in the present invention or the prior art, a brief introduction will be given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0042] Figure 1 A schematic flow chart of a method for visualizing expression profiles of multiple tissues and single cells in rice provided by the present invention;

[0043] Figure 2 This is a schematic diagram of the structure of a visualization analysis system for rice multi-tissue single-cell expression profiles provided by the present invention. DETAILED DESCRIPTION

[0044] In order to make the purpose, technical solutions and advantages of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below in conjunction with the drawings in the present invention. Obviously, the embodiments described are part of the embodiments of the present invention, not all of the embodiments, and they should not be understood as limitations on the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention. In the description of the present invention, it should be understood that the terms used are only for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0045] like Figure 1 As shown, the present invention provides a method for visual analysis of rice multi-tissue single-cell expression profiles, comprising:

[0046] S1: The original single-cell transcriptome sequencing data of rice tissues were aligned and processed to obtain a standardized transcriptome dataset.

[0047] Wherein, in step S1, the rice tissues include: rice crown roots, rice stems, rice leaves, rice flag leaves, rice stem tips, rice tiller buds, rice panicles, and rice seeds.

[0048] The present invention first uses Cell Ranger to compare and process the original scRNA-seq data, which involves complex sequence alignment algorithms and data preprocessing steps. In the data processing of rice tissue, the present invention uses Cell Ranger to perform sequence alignment based on the MSUv7.0 reference genome. During the alignment process, the software calculates the similarity score of each read with the reference sequence, and only reads with a similarity exceeding a preset threshold are retained. Specifically, for the other seven tissues of rice, such as rice stalks, leaves, flag leaves, stem tips, tiller buds, panicles, and seeds, the same alignment process is used for processing. After the alignment is completed, Cell Ranger counts the number of reads of each gene in each cell to generate a gene expression count matrix. The rows of the matrix represent genes, the columns represent cells, and the value of each element represents the number of expression reads of the gene in the cell. Finally, after quality control and filtering, the original scRNA-seq data of the eight tissues are converted into a standardized gene expression count matrix format to constitute a standardized transcriptome data set.

[0049] S2: The normalized transcriptome dataset is corrected for batch effects and integrated to obtain a whole-plant expression matrix.

[0050] Wherein, step S2 further includes:

[0051] S21: performing normalization processing on the gene expression data in the standardized transcriptome dataset to obtain normalized expression data.

[0052] Furthermore, in step S21, the single-cell transcriptome data are first received and normalized. Specifically, the normalization process of the present invention is performed using the Seurat (v4.3) software package, including first performing a logarithmic transformation on the original gene expression count matrix to convert it from a Poisson distribution to an approximate normal distribution, reducing the weight of highly expressed genes, and avoiding a small number of highly expressed genes dominating subsequent analysis; then performing Z-score normalization, calculating the expression value of each gene minus the mean of the gene, and then dividing by the standard deviation. Z-score normalization ensures that the expression values ​​of different genes have the same numerical range, eliminating the impact of differences in expression levels between genes on downstream analysis.

[0053] S22: Perform multi-tissue integration on the normalized expression data based on the Harmony algorithm to establish a whole-plant expression matrix.

[0054] Furthermore, in step S22, the present invention uses the Harmony algorithm to process the integration problem of multiple batches of single-cell data to eliminate non-biological differences caused by different experimental batches, tissue sources or treatment conditions. The batch effect mainly comes from technical differences between tissues and changes in experimental conditions.

[0055] Specifically, the present invention uses the Harmony algorithm to calculate the proportion of cells from different tissues within each cluster. If the proportion of cells from a particular tissue within a cluster is too high, the algorithm adjusts the centroid of that cluster, moving it away from the center of cell distribution for that tissue. This algorithm repeats 10 iterations until the cluster assignments stabilize, ultimately outputting the corrected low-dimensional embedding coordinates. These corrected 50-dimensional embedding coordinates are then combined with the original hypervariable gene expression matrix to obtain the full-plant expression matrix.

[0056] S3: performing cell type annotation on the whole plant expression matrix to obtain a cell type annotation system.

[0057] Wherein, step S3 further includes:

[0058] S31: Construct a marker gene database based on the collected rice marker genes and homologous marker genes.

[0059] Furthermore, the marker gene database is constructed based on two data sources: literature mining and homologous gene analysis. Literature mining systematically searches for relevant literature on rice single-cell transcriptomics and collects verified tissue-specific marker genes. In the implementation of the present invention, the search scope includes PubMed, Web of Science and other databases, and extracted marker gene information from the retrieved literature, including gene name, MSU gene ID, RAP gene ID, marked cell type, verification method, etc.

[0060] Homologous marker gene analysis uses known cell type marker genes of model plants to identify homologous genes in rice through sequence alignment. The sequence alignment uses the BLAST algorithm, and the alignment results are manually screened to eliminate homologous gene pairs with inconsistent functional annotations.

[0061] The marker gene database ultimately contains 312 marker genes for 15 major cell types, covering eight root cell types (epidermal cells, cortical cells, endodermal cells, pericycle cells, xylem cells, phloem cells, meristem cells, and root cap cells), four leaf cell types (mesophyll cells, vascular cells, epidermal cells, and guard cells), and three stem and other tissue cell types (vascular cells, parenchyma cells, and epidermal cells). The database is stored in JSON format and includes fields such as gene ID, gene name, cell type, and expression specificity score.

[0062] S32: Based on the marker gene database, the cells in the whole plant expression matrix are annotated by type through in sit experimental verification to obtain a cell type annotation system.

[0063] Furthermore, in step S32 of the present invention, 10 representative marker genes were selected for RNA in situ hybridization (ISH) validation, verifying their expression patterns in specific tissues and cell types. The experimental materials were rice root, leaf, and stem tissues, sampled 14 days after sowing. ISH used a digoxigenin-labeled antisense RNA probe at 65°C for 16 hours, and developed using an NBT / BCIP substrate for 2-6 hours. The experimental results achieved 92% consistency with the single-cell transcriptome data, validating the accuracy of the marker gene database.

[0064] The cell type annotation system ultimately established a classification system that includes 15 cell types. Each cell type contains a list of specific marker genes, expression feature descriptions, tissue distribution information, etc. The annotation results are indexed by cell barcodes and store each cell's type label, confidence score, candidate type and other information.

[0065] S4: The whole-plant expression matrix of the fusion cell type annotation system was visualized and dimensionality reduced, and a co-expression network was constructed to obtain a modular tissue association analysis model.

[0066] Wherein, step S4 further includes:

[0067] S41: Mapping the whole plant expression matrix to a two-dimensional space to obtain a two-dimensional space representation.

[0068] Furthermore, in step S42 of the present invention, the whole-plant expression matrix obtained above is mapped to a two-dimensional space using the Unified Mapping (UMAP) dimensionality reduction algorithm. The UMAP algorithm first constructs a k-nearest neighbor graph in the high-dimensional space, calculates local connectivity weights, and then performs a probabilistic simplification process to convert the connectivity weights in the high-dimensional space into a probability distribution. Subsequently, the two-dimensional space is initialized based on the probability distribution, and spectral embedding is performed, ultimately outputting a two-dimensional spatial representation represented by a coordinate matrix.

[0069] S42: Fusing the two-dimensional spatial representation with the standardized cell type annotation system to generate a cell cluster distribution map.

[0070] Furthermore, the cell cluster distribution map is generated using the SVG vector graphics format, which supports lossless scaling and interactive operations. SVG graphic elements include four components: scatter points, color coding, legend, and coordinate axes. The legend component also includes the cell type name, corresponding color, and cell number statistics.

[0071] It should also be noted that the cell cluster distribution map data structure of the present invention also includes an interactive function interface that supports click selection, region zooming, and type filtering operations. The click selection function is implemented through mouse event monitoring, and when a scatter point is clicked, detailed cell information, including cell barcode, coordinates, type, and confidence level, is displayed. The region zoom function is implemented by dragging and selecting, and the view range is automatically adjusted after selecting a rectangular area. The type filtering function is controlled by a checkbox, and only the corresponding cells are displayed when a specific cell type is checked.

[0072] S43: Calculating the correlation between genes in the whole-plant expression matrix to obtain a gene correlation matrix.

[0073] Furthermore, the correlation calculation process first constructs a gene expression matrix, with rows representing genes and columns representing cells. Each element in the matrix represents the expression value of a specific gene in a specific cell. Next, the mean expression value and standard deviation of each gene are calculated. The correlation calculation between gene pairs is optimized using matrix operations to avoid the inefficiency of pair-by-pair calculations. Specifically, the gene expression matrix is ​​normalized, and the expression value of each gene is subtracted from the mean and divided by the standard deviation. The correlation matrix is ​​then calculated using matrix multiplication.

[0074] S44: Based on the cell cluster distribution map and the gene correlation matrix, a module tissue correlation analysis model is constructed.

[0075] Furthermore, the modular organization association analysis model was constructed using the hdWGCNA (high-dimensional Weighted Gene Co-expression Network Analysis) method. The hdWGCNA algorithm first constructs a weighted co-expression network based on the gene correlation matrix. During network construction, the correlation matrix is ​​first converted into an adjacency matrix, which is then further converted into a topological overlap matrix. The topological overlap matrix accounts for indirect connections between genes and improves network stability. Each element in the matrix represents the topological overlap between two genes.

[0076] Then, the gene module is identified by using the hierarchical clustering method, the clustering method is the average linkage method, the hierarchical clustering generates a gene dendrogram, and the dendrogram is divided into different modules by the dynamic cutting method.

[0077] Module-tissue association analysis calculated the expression patterns of each module across eight tissues. Module expression values ​​were calculated as the first principal component of all gene expression values ​​within the module. Principal component analysis was performed using singular value decomposition. The dimensions of the principal component loading matrix represent the contribution weight of each gene to module expression. The rows of the module expression matrix represent modules, and the columns represent tissues.

[0078] The tissue-specific score calculates the enrichment of each module in a specific tissue. A positive value in the tissue-specific score matrix indicates that the module is highly expressed in a specific tissue, while a negative value indicates low expression.

[0079] The final output of the module-tissue association analysis model consists of three components: gene module information, a module expression matrix, and a tissue-specific scoring matrix. Gene module information includes the module number, module color, and module size to which each gene belongs. The module expression matrix records the expression pattern of each module in different tissues. The tissue-specific scoring matrix quantifies the strength of the association between the module and the tissue.

[0080] S5: An interactive terminal is established based on the module organization correlation analysis model, and data visualization analysis is achieved through the interactive terminal.

[0081] Furthermore, the present invention constructs a server system in step S5 for interaction. The interactive end is established using a B / S architecture, the front end is built based on the React framework, and the back end deploys static resource services through Apache / 2.4.63. The front-end architecture includes a three-level navigation system: a main menu layer, an organizational map portal layer, and a gene search panel layer. The main menu layer includes three main entrances: Home, Tools, and Download, and each entrance corresponds to a different functional module. The organizational map portal layer includes UMAP map entrances for eight organizations, and each entrance loads the cell cluster distribution map of the corresponding organization. The gene search panel layer includes an organization selector and a gene ID input box, and supports MSU / RAP gene ID format input.

[0082] Among them, also include:

[0083] Step S6: Based on the interactive end, the root cell type is predicted for the input expression matrix through the pre-trained XGBoost model to obtain the root cell probability prediction result.

[0084] Wherein, step S6 further includes:

[0085] S61: Receive input expression matrix.

[0086] The input expression matrix is ​​then validated and converted to a Python pandas DataFrame format, with rows representing genes and columns representing cells. The DataFrame index is the gene ID, the columns are the cell barcodes, and the matrix elements are the gene expression counts in the corresponding cells. During this conversion, the sparse matrix format is converted to a dense matrix format for easier processing.

[0087] S62: Preprocessing the input expression matrix using a scanpy standardization algorithm to obtain preprocessed expression data.

[0088] Furthermore, in step S62, the scanpy normalization algorithm of the present invention performs preprocessing on the received matrix, including quality control, normalization, and high-variable genes. The quality control process calculates cell and gene quality indicators, including the number of genes per cell, the number of cells per gene, and the proportion of mitochondrial genes, and then screens them based on thresholds. The normalization process uses the scanpy.pp.normalize_total function to normalize the total expression level of each cell to 10,000. The scanpy.pp.highly_variable_genes function is used to identify high-variable genes based on expression mean and dispersion.

[0089] The preprocessed expression data is output as a standardized gene expression matrix with the dimension of the number of highly variable genes × the number of cells. The matrix elements are logarithmically normalized expression values ​​with a numerical range of 0-10. At the same time, the preprocessing process also outputs auxiliary information such as quality control indicators, a list of highly variable genes, and normalization parameters.

[0090] S63: Using the pre-trained XGBoost model, cell type classification is performed based on the highly variable gene features in the pre-processed expression data to obtain a root cell probability prediction result.

[0091] Furthermore, the XGBoost model was pre-trained based on single-cell transcriptome data from rice roots. The training set contained 25,000 labeled cells, covering nine root cell types: epidermal cells, cortical cells, endodermal cells, pericycle cells, xylem cells, phloem cells, meristem cells, root cap cells, and undifferentiated cells.

[0092] The cell type classification process feeds preprocessed expression data into a trained XGBoost model. The input data format is a cell × gene matrix, with each row representing a cell and each column representing the expression value of a highly variable gene. The model output is a cell × type probability matrix, with each row representing a cell and each column indicating the probability of belonging to a specific cell type. Probability values ​​range from 0 to 1, and the sum of probabilities in each row is 1. The root cell probability prediction results contain four fields: cell barcode, predicted type, probability distribution, and confidence score. The predicted type is the cell type with the highest probability, the probability distribution is the probability values ​​of the nine cell types, and the confidence score is the difference between the highest and second highest probability. The confidence score ranges from 0 to 1, with higher values ​​indicating more reliable predictions.

[0093] like Figure 2 As shown, the present invention also provides a visualization analysis system for rice multi-tissue single-cell expression profiles, comprising:

[0094] The front-end interactive interface is a responsive front-end built on the React framework, including a main menu, an organizational map portal, and a gene search panel with a three-tiered navigation architecture;

[0095] Backend service architecture, which is a static resource server deployed through Apache / 2.4.63 to support high-throughput access and data transmission;

[0096] A tissue map browsing module, comprising an input layer, a visualization engine, and an output layer, for displaying and processing multi-dimensional tissue cell clustering maps based on the UMAP dimensionality reduction algorithm;

[0097] A gene retrieval module, comprising an input layer, an analysis engine, and an output layer, for retrieving and analyzing the cross-tissue expression patterns of target genes based on MSU / RAP gene IDs;

[0098] Network analysis module: used to construct single-cell co-expression networks based on the hdWGCNA algorithm and generate module tissue correlation analysis models.

[0099] Among them, also include:

[0100] The RSCA annotation tool module is built based on a pre-trained XGBoost model and is used to automatically annotate root cell types on 10x Genomics expression matrices uploaded by users.

[0101] Among them, the user selection modes supported by the input layer of the tissue map browsing module include specific tissue and whole plant views. The visualization engine of the tissue map browsing module supports scaled UMAP dimensionality reduction maps based on SVG vector graphics to display cell cluster distribution. The output layer of the tissue map browsing module provides export and metadata download functions.

[0102] The input layer of the gene retrieval module supports gene ID input and tissue screening. The analysis engine of the gene retrieval module quantifies the TPM expression intensity of the target gene in the selected cell type through violin plots and generates a cross-tissue expression pattern heat map. The output layer of the gene retrieval module displays single-cell expression data through interactive charts.

[0103] The network analysis module constructs a single-cell co-expression network based on preset soft threshold parameters, displays the core regulatory modules through a regulatory network diagram, annotates related genes based on module organization association analysis, and supports the export of network format files.

[0104] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, i.e., they may be located in one location or distributed across multiple network units. Some or all of the modules may be selected based on actual needs to achieve the objectives of the present embodiment. Persons of ordinary skill in the art will be able to understand and implement the present invention without inventive effort.

[0105] Furthermore, the present invention provides a visualization analysis system for rice multi-tissue single-cell expression profiles, which is implemented as follows.

[0106] This system adopts B / S architecture (Browser / Server Architecture). The front-end is built with React to build interactive interface, and the back-end is deployed with static resource service through Apache / 2.4.63. It includes three core modules:

[0107] Tissue map browsing module. Input layer: supports user selection of specific tissues (crown roots, stems, leaves, etc.) or whole plant views; visualization engine: UMAP dimensionality reduction plots display cell clustering and type distribution (SVG vector images support scaling); tissue details page annotates cell type composition and key marker genes (such as the root endodermis marker OsXTH21); output layer: provides PNG / SVG format export and metadata download.

[0108] Gene search module. Input layer: supports MSU / RAP gene IDs (e.g., LOC_Os01g01010) and tissue filtering; Analysis engine: Violin plots quantify the expression intensity (TPM value) of target genes in selected cell types, and heat maps compare gene expression patterns across eight tissues; Output layer: Interactive chart linkage, click on UMAP data points to display single-cell expression levels.

[0109] Network Analysis Module. Constructs single-cell co-expression networks based on hdWGCNA (soft_power = 9). Visualization features: Regulatory network diagrams display core regulatory modules, module-tissue association analysis, and annotation of genes related to roots and leaves. Data interface: Supports export of network files in network format.

[0110] RSCA annotation tool. Algorithm architecture: Based on a pre-trained XGBoost model; Input layer: User-uploaded 10xGenomics-formatted expression matrix; Feature processing layer: Scanpy normalization (scanpy.pp.log1p) and screening of highly variable genes; Classification layer: Probability prediction of nine root cell types; Online deployment: Independent subpage, with sample data provided for testing.

[0111] For the web server implementation, it includes: a responsive front-end; a three-level navigation architecture: a main menu (Home / Tools / Download, etc.), an organization map portal (eight organization UMAP entries), and a gene search panel (organization selection + gene ID input).

[0112] Visualization panel: integrated with JBrowse genome browser.

[0113] High-performance service architecture: Load optimization: All genes are visualized locally, UMAP images are pre-rendered, and the average response time is <800ms.

[0114] Data open ecosystem: The download center provides standardized data packages (cell expression matrix / annotation file / network topology), and GitHub open source RSCA core code.

[0115] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, or of course, by hardware. Based on this understanding, the essence of the above technical solution or the part that contributes to the existing technology can be embodied in the form of a software product. The computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a magnetic disk, an optical disk, etc., and includes a number of instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or certain parts of the embodiments.

[0116] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A method for visual analysis of rice multi-tissue single-cell expression profiles, characterized in that: include: S1: Align the sequencing data of the original single-cell transcriptome of rice tissue to obtain a standardized transcriptome dataset; S2: performing batch effect correction on the normalized transcriptome dataset and integrating the dataset to obtain a whole-plant expression matrix; S3: performing cell type annotation on the whole plant expression matrix to obtain a cell type annotation system; S4: Visualize and reduce the dimensionality of the whole-plant expression matrix integrated with the cell type annotation system, and construct a co-expression network to obtain a modular tissue correlation analysis model; S5: An interactive terminal is established based on the module organization correlation analysis model, and data visualization analysis is achieved through the interactive terminal.

2. The method for visual analysis of rice multi-tissue single-cell expression profiles according to claim 1, characterized in that: In step S1, the rice tissues include: rice crown roots, rice stems, rice leaves, rice flag leaves, rice stem tips, rice tiller buds, rice panicles, and rice seeds.

3. The method for visual analysis of rice multi-tissue single-cell expression profiles according to claim 1, characterized in that: Step S2 further comprises: S21: performing normalization processing on the gene expression data in the standardized transcriptome dataset to obtain normalized expression data; S22: Perform multi-tissue integration on the normalized expression data based on the Harmony algorithm to establish a whole-plant expression matrix.

4. The method for visual analysis of rice multi-tissue single-cell expression profiles according to claim 1, characterized in that: Step S3 further comprises: S31: Construct a marker gene database based on the collected rice marker genes and homologous marker genes; S32: Based on the marker gene database, the cells in the whole plant expression matrix are annotated by type through in sit experimental verification to obtain a cell type annotation system.

5. The method for visual analysis of rice multi-tissue single-cell expression profiles according to claim 1, characterized in that: Step S4 further comprises: S41: mapping the whole plant expression matrix to a two-dimensional space to obtain a two-dimensional space representation; S42: fusing the two-dimensional spatial representation with the standardized cell type annotation system to generate a cell cluster distribution map; S43: calculating the correlation between genes in the whole plant expression matrix to obtain a gene correlation matrix; S43: Based on the cell cluster distribution map and the gene correlation matrix, a module tissue correlation analysis model is constructed.

6. The method for visual analysis of rice multi-tissue single-cell expression profiles according to claim 1, characterized in that: Also includes: Step S6: Based on the interactive end, the root cell type is predicted for the input expression matrix through the pre-trained XGBoost model to obtain the root cell probability prediction result.

7. A method for visualizing and analyzing rice multi-tissue single-cell expression profiles according to claim 6, characterized in that: Step S6 further comprises: S61: receiving an input expression matrix; S62: preprocessing the input expression matrix using a scanpy normalization algorithm to obtain preprocessed expression data; S63: Using the pre-trained XGBoost model, cell type classification is performed based on the highly variable gene features in the pre-processed expression data to obtain a root cell probability prediction result.

8. A visualization analysis system for rice multi-tissue single-cell expression profiles, characterized by: include: The front-end interactive interface is a responsive front-end built on the React framework, including a main menu, an organizational map portal, and a gene search panel with a three-tiered navigation architecture; Backend service architecture, which is a static resource server deployed through Apache / 2.4.63 to support high-throughput access and data transmission; A tissue map browsing module, comprising an input layer, a visualization engine, and an output layer, for displaying and processing multi-dimensional tissue cell clustering maps based on the UMAP dimensionality reduction algorithm; A gene retrieval module, comprising an input layer, an analysis engine, and an output layer, for retrieving and analyzing the cross-tissue expression patterns of target genes based on MSU / RAP gene IDs; Network analysis module: used to construct single-cell co-expression networks based on the hdWGCNA algorithm and generate module tissue correlation analysis models.

9. The visualization analysis system for rice multi-tissue single-cell expression profiles according to claim 8, characterized in that: Also includes: The RSCA annotation tool module is built based on a pre-trained XGBoost model and is used to automatically annotate root cell types on 10x Genomics expression matrices uploaded by users.

10. The visualization analysis system for rice multi-tissue single-cell expression profiles according to claim 8, characterized in that: The input layer of the tissue map browsing module supports user selection modes including specific tissue and whole plant views. The visualization engine of the tissue map browsing module supports scaled UMAP dimensionality reduction maps based on SVG vector graphics to display cell cluster distribution. The output layer of the tissue map browsing module provides export and metadata download functions. The input layer of the gene retrieval module supports gene ID input and tissue screening. The analysis engine of the gene retrieval module quantifies the TPM expression intensity of the target gene in the selected cell type through violin plots and generates a cross-tissue expression pattern heat map. The output layer of the gene retrieval module displays the single-cell expression data through interactive charts. The network analysis module constructs a single-cell co-expression network based on preset soft threshold parameters, displays the core regulatory modules through a regulatory network diagram, annotates related genes based on module organization association analysis, and supports the export of network format files.

Images