Single-cell multi-omics data analysis system construction method based on containerization technology

Abstract

The invention discloses a construction method of a single-cell multi-omics data analysis system based on a containerization technology, and the construction method of the single-cell multi-omics data analysis system based on the containerization technology comprises the following steps: constructing a single-cell multi-omics analysis modular toolkit; a production mirror image is constructed on the high-performance cluster through a Single container engine, and deployment of a production environment is completed; a single-cell multi-omics automatic analysis platform is built through a snakemake process engine; and a JupyterLab interactive analysis platform for exploratory analysis of the single-cell multi-omics data is constructed. According to the single-cell multi-omics data analysis system construction method based on the containerization technology, the data acquisition link in the analysis process is accelerated, an automatic working process is formed to solve the problem of complex process, and the actual requirements of multiple users on a high-performance computing cluster for scientific exploration and analysis of single-cell data are met.

Description

Technical Field

[0001] This invention relates to the field of single-cell transcriptome data processing technology, and in particular to a method for constructing a containerized single-cell multi-omics data analysis system. Background Technology

[0002] With the advent of single-cell transcriptome sequencing technology, single-cell omics sequencing technology has developed rapidly in recent years, with various single-cell technologies, methods, and platforms emerging quickly (such as single-cell transcriptome sequencing, single-cell immune repertoire, and single-cell multimodal sequencing), providing significant convenience for scientific research in the life sciences field. However, due to the differences between various technologies and platforms, there is a diversity in data formats and data processing methods, which also brings many challenges to single-cell transcriptome data processing.

[0003] Single-cell transcriptome sequencing technology has developed rapidly in recent years. Its core goal is to analyze gene expression at the single-cell level. In terms of throughput and omics integration technology, it has evolved from single transcriptome sequencing to multi-omics combined sequencing; in terms of analytical depth and analytical capability, it has evolved from static maps to dynamic and causal inference; and in terms of data integration and model building technology, it has evolved from single data analysis to multimodal integration and basic model development.

[0004] However, single-cell transcriptome sequencing technology also faces many challenges, mainly in the following four aspects: First, due to the diversity of platforms and technologies, the formats of single-cell transcriptome data vary considerably. Data preprocessing often requires manual intervention, resulting in long processing cycles and the risk of errors. Second, single-cell transcriptome data processing methods are developing rapidly, and software is frequently updated. To maintain the novelty and accuracy of the analysis methods, regular version iterations are necessary. Therefore, it is urgent to build an efficient, highly integrated, modular, and reproducible analysis toolkit. Third, single-cell transcriptome data processing consumes significant computing resources, requiring the use of high-performance computing (HPC) clusters. However, the currently popular Docker technology in HPC clusters has some limitations, such as the lack of root privileges for regular developers and resource management (memory, CPU) requirements. Finally, single-cell sequencing data analysis is usually a scientific research exploration, requiring high real-time performance and interactivity in the data processing process.

[0005] Therefore, there is an urgent need for a method to construct a single-cell multi-omics data analysis system to solve the problem of organizing source data from various single-cell technology platforms. This method should accelerate the data acquisition stage in the analysis process by generating a unified data storage medium; solve the problems of deployment and maintenance difficulties and complexities caused by multiple software dependencies and inconsistent version numbers in traditional technical solutions; form an automated workflow to solve the problem of process complexity; and simultaneously meet the actual needs of multiple users on high-performance computing clusters to conduct scientific exploration and analysis of single-cell data. Summary of the Invention

[0006] In view of the above-mentioned shortcomings of current single-cell transcriptome sequencing technology, this invention provides a method for constructing a containerized single-cell multi-omics data analysis system, which can ensure the traceability of the single-cell transcriptome data analysis process and the reproducibility of data analysis results; solve the problems of high deployment and maintenance difficulty and complexity caused by multiple dependent software and chaotic version numbers in traditional technical solutions; realize the automated generation of basic data analysis reports; and meet the actual needs of multiple users on high-performance computing clusters to conduct scientific exploration and analysis of single-cell data.

[0007] To achieve the above objectives, the embodiments of the present invention adopt the following technical solutions:

[0008] A method for constructing a single-cell multi-omics data analysis system using containerization technology, characterized in that the method includes the following steps:

[0009] Develop a modular toolkit for single-cell multi-omics analysis;

[0010] The production environment is deployed by building production images on a high-performance cluster using the Singularity container engine.

[0011] A single-cell multi-omics automated analysis platform was built using the snakemake workflow engine.

[0012] Develop an interactive JupyterLab analysis platform for exploratory analysis of single-cell multi-omics data;

[0013] Modular toolkits, automated analysis platforms, and interactive analysis platforms are used to build a single-cell multi-omics data analysis system through containerization and process management technologies.

[0014] According to one aspect of the invention, the modular toolkit for single-cell multi-omics analysis includes eight modular tools.

[0015] According to one aspect of the present invention, the eight modular tools include: a data preprocessing and quality control module; a data integration and batch correction module; a dimensionality reduction, clustering and cell annotation module; a multi-omics data integration and joint analysis module; a trajectory inference and cell dynamics module; a gene regulatory network analysis module; a functional enrichment analysis module; and an intercellular communication analysis module.

[0016] According to one aspect of the present invention, the modular tool is compiled using the R language and can be invoked via command line in a Linux environment or by loading an R language package in the R kernel of the JupyterLab interactive analysis platform; the modular toolkit can quickly realize the creation, updating, extraction, downsampling, and conversion operations of data storage media by calling subroutine modules.

[0017] According to one aspect of the invention, the modular toolkit can create data storage media based on processing data from different single-cell platform sources; the update refers to updating the existing data storage media by adding new feature data; the extraction refers to extracting sub-data from the existing data storage media according to analysis requirements, that is, extracting data related to the analysis requirements from the existing data; the downsampling refers to randomly sampling the existing data storage media to reduce the amount of data without affecting the analysis results; and the conversion refers to converting the format of the data storage media according to analysis requirements.

[0018] According to one aspect of the present invention, the modular toolkit for single-cell multi-omics analysis is constructed using a development directory, which contains multiple folders, including: an R folder, an exec folder, a man folder, and a data folder.

[0019] According to one aspect of the present invention, the R folder is used to store general basic and functional functions for single-cell multi-omics analysis; the exec folder is used to store executable files, including the main program module files sctool and scVis and their respective subroutine modules; the man folder is used to store R markdown documents explaining the use of functions in R; and the data folder is used to store the data sets involved in the processing of each module in the toolkit.

[0020] According to one aspect of the present invention, the step of building a production image on a high-performance cluster using the Singularity container engine to complete the deployment of the production environment includes the following steps:

[0021] Install the Singularity container engine on the host where the image was created;

[0022] Write a Singularity image definition file, which predefines the base image used to build the image and the commands executed during the build process;

[0023] The Singularity container engine uses the `build` command to read the Singularity image definition file and build the binary image file.

[0024] On the cluster where the production environment needs to be deployed, copy the binary image file created above to complete the deployment of the production environment.

[0025] According to one aspect of the present invention, the construction of a single-cell multi-omics automated analysis platform using the snakemake workflow engine includes the following steps:

[0026] Install the snakemake workflow engine;

[0027] Based on the pre-built modular toolkit for single-cell multi-omics analysis, the Snakemake automation process file is written using the Sankemake syntax to define the standard single-cell multi-omics analysis tasks to be performed.

[0028] Pull the complete Snakemake automation workflow files, and after completing the environment initialization configuration and project basic information configuration, obtain the single-cell multi-omics automated analysis platform.

[0029] According to one aspect of the present invention, the construction of the JupyterLab interactive analysis platform for exploratory analysis of single-cell multi-omics data includes automatically converting source code into a JupyterLab executable file notebook through the plugin jupytext, maintaining consistency between the exploratory data analysis method and the automated analysis workflow code.

[0030] Advantages of this invention: Through the above technical solution, this invention...

[0031] (1) Based on the modular tool for single-cell transcriptome analysis, the problem of source data processing for various single-cell technology platforms is solved. By generating a unified data storage medium, the data acquisition stage in the analysis process is accelerated, and various operations on the data storage medium are supported. The modular toolkit has a unified interface, and the functions of each analysis module can be implemented independently through sub-commands, which facilitates the further development and expansion of the analysis tool. At the same time, the toolkit realizes the real-time saving of analysis scripts, running parameters, environment configuration parameters, running logs, etc., ensuring the traceability of the single-cell transcriptome data analysis process and the reproducibility of data analysis results.

[0032] (2) Based on Singularity container technology, it solves the problems of large deployment and maintenance difficulty and complexity caused by many dependent software and chaotic version numbers in traditional technical solutions: On the one hand, Singularity is compatible with Docker images, which is convenient to build, and the built images can be directly copied and ported to HPC clusters with Singularity installed for reuse, which saves environment configuration costs and ensures the reproducibility of data analysis results; On the other hand, Singularity supports running on systems without root privileges, which takes into account data security and convenience, and is more in line with the real scenario of single-cell bioinformatics analysis.

[0033] (3) Based on the Snakemake workflow engine technology, modular tools are combined to form an automated workflow for single-cell transcriptome data. It can be used with high-performance computing cluster task scheduling system to realize the automated generation of basic data analysis reports. It also supports Singularity environment image, which has a high advantage in workflow upgrade, optimization and migration.

[0034] (4) By enabling the JupyterLab web service system, the actual needs of multiple users on the high-performance computing cluster to conduct scientific exploration and analysis of single-cell data were met. Attached Figure Description

[0035] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0036] Figure 1 This is a flowchart illustrating a construction method for building a single-cell multi-omics analysis system based on the Snakemake workflow engine, JupyterLab web service, and Singularity container engine, as described in this invention.

[0037] Figure 2 This is a simplified diagram of the architecture of a single-cell multi-omics analysis system built based on the Snakemake workflow engine, JupyterLab web service, and Singularity container system, as described in this invention.

[0038] Figure 3 This is a schematic diagram illustrating the operation of a functional analysis tool based on multi-platform single-cell data as described in this invention.

[0039] Figure 4This is a schematic diagram illustrating the process of constructing and operating a single-cell transcriptome data storage medium according to the present invention.

[0040] Figure 5 This is a flowchart illustrating a functional analysis module based on multi-platform single-cell data as described in this invention.

[0041] Figure 6 This is a schematic diagram illustrating the process of building a high-performance cluster production environment based on Singularity, as described in this invention. Detailed Implementation

[0042] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0043] Example 1

[0044] This invention proposes a method for constructing a single-cell transcriptome data analysis system based on Singularity containerization technology, the specific steps of which include:

[0045] (1) Construction of basic toolkit: Construction of a modular toolkit for single-cell transcriptome analysis;

[0046] (2) Production environment deployment: Build production images on a high-performance cluster using the Singularity container engine;

[0047] (3) Building an automated analysis platform: A single-cell transcriptome automated analysis platform was built using the snakemake workflow engine;

[0048] (4) Construction of interactive analysis platform: Construct JupyterLab interactive analysis platform for exploratory analysis of single-cell transcriptome data.

[0049] Example 2

[0050] Development and Construction of Modular Tools for Single-Cell Transcriptome Analysis

[0051] The modular tool for single-cell transcriptome analysis is a software program compiled in R. It can be called via command line in a Linux environment to realize functions such as the construction and operation of single-cell transcriptome data storage media, data preprocessing, functional analysis and visualization of analysis results. It can also be called by loading R packages in the R kernel of the JupyterLab interactive analysis platform to perform exploratory analysis.

[0052] The steps for developing and constructing a modular tool for single-cell transcriptome analysis are as follows:

[0053] (1) Develop a modular tool for single-cell transcriptome analysis according to the development directory, including: R folder, exec folder, man folder, and data folder.

[0054] The R folder contains general-purpose basic and functional files for single-cell transcriptome analysis, such as the `save_ggplot` function for saving visualization results; and the `RunDimReduc` function for dimensionality reduction of data storage media, supporting various dimensionality reduction algorithms in single-cell data analysis fields, including PCA (Principal Component Analysis), LSI (Latent Semantic Indexing), CCA (Canonical Correlation Analysis), MNN (Mutual Nearest Neighbors), and Harmony (a single-cell transcriptome integration algorithm).

[0055] The `exec` folder is used to store executable files, containing the main program module files `sctool` and `scVis`, and their corresponding subroutine modules. The main program module files `sctool` and `scVis` are toolkit calling interfaces, allowing commands to be executed via command line in a Linux environment. These subroutine modules primarily implement specific functional analysis and visualization programs. The `sctool` interface is mainly used for calling functional analysis tools, while the `scVis` interface is mainly used for calling result visualization tools. The subroutine module name corresponding to `sctool` must be in the format `utils-"submodule name"-".R"`, and this subroutine module is mainly used to implement... Figure 4 Functional analysis in scVis, such as dimensionality reduction clustering, cell type annotation, copy number variation analysis, etc.; the corresponding subroutine module name in scVis should be plot-“submodule name”-“.R”, and this subroutine module is mainly used to implement Figure 4 The visualization of analysis results includes heatmaps, violin plots, bar charts, spatial visualizations, MA plots, and scatter plots.

[0056] The man folder is used to store Rmarkdown documentation explaining the usage of functions in R. It can be automatically generated based on the function comments in R during the development process.

[0057] The data folder is used to store the data sets involved in the processing of each module in the toolkit.

[0058] (2) The main program module of the modular toolkit for single-cell multi-omics analysis was developed and constructed to enable the configuration of common general parameters. Common general parameters include input file specification parameter --input, data format specification parameter --formart, output directory specification parameter --output, etc. These parameters can be reused in relevant subroutine modules. At the same time, general functions were configured and reused in relevant subroutine modules. The configuration of general functions involves operations such as parsing running parameters, setting multi-threading, collecting analysis environment information, printing running logs, and copying source code. Through the above operations, a running log file named with a timestamp (implemented by futil.logger::*flog.appender*) is provided in the log folder of each analysis result directory, thereby ensuring the controllability of the analysis process. In addition, each time the program runs successfully, a session_info file containing parameter setting information, running platform, and software version information (implemented by sessioninfo::*session_info*) and a code file (implemented by funr::*get_script_path function*) named with a timestamp are provided, thereby ensuring the reproducibility of the analysis results.

[0059] (3) Quickly install the modular toolkit for single-cell multi-omics analysis into the production environment for use. After development is completed, use the Git version control system to mark the version, and then use the common R package installation method to install the modular toolkit for single-cell multi-omics analysis. After that, it can be called uniformly in the command line through the main program module.

[0060] The specific usage instructions for the modular toolkit for single-cell multi-omics analysis are as follows:

[0061] 1) You can execute the main program module file sctool or scVis via the command line, such as: sctool --help, to query the main program parameter descriptions and search for supported subroutine modules;

[0062] 2) By executing the main program and the key fields of the subroutine module through the command line, such as: sctool create—help, you can query the parameters and running examples of the subroutine module for the corresponding data medium.

[0063] (4) The operation diagram of a functional analysis tool based on multi-platform single-cell data, which uses command-line instructions, is shown below. Figure 3 As shown, subroutine modules can be invoked to quickly perform operations related to the data storage medium. Its specific functions include:

[0064] It supports the creation of data storage media for processing data from different single-cell platforms. It supports input data formats such as 10x Genomics platform mtx format data, 10x Genomics platform hdf5 format data, BD Rhapsody platform text format data or other text formats (such as csv, tsv), etc. It supports output data formats such as rds and h5seurat formats supported by the Serurat ecosystem.

[0065] It can perform batch and fast conversion between various mainstream data structures, and supports mutual conversion between data formats such as Seurat, Monocle2, Monocle3, SingleCellExperiment, and Scanpy, to meet the data structures supported by different single-cell analysis software;

[0066] It can quickly realize the updating, extraction, downsampling, and conversion of data storage media.

[0067] Updating refers to updating existing data storage media by adding new feature data;

[0068] Extraction refers to extracting sub-data from existing data storage media according to analysis needs, that is, extracting data related to analysis needs from existing data;

[0069] Downsampling refers to the random sampling of existing data storage media based on specified groups, such as groups or clusters, when the number of cells is particularly large, in order to reduce the amount of data without affecting the analysis results and thus save computational costs.

[0070] Conversion refers to the conversion of data storage media formats according to analysis needs. Currently, it supports conversion between formats such as monocle2, monocle3, h5Seurat, SingleCellExperiment, Scanpy, and anndata.

[0071] (5) A schematic diagram of the workflow for constructing and manipulating single-cell transcriptome data storage media using command-line instructions is shown below. Figure 4 As shown, the process of calling the subroutine module includes the following:

[0072] A data storage medium containing single-cell transcriptome, single-cell immune repertoire, and single-cell multimodal data was obtained, the data of which came from platforms such as 10x Genomics, BD Rhapsody, and InDrop;

[0073] The acquired data undergoes preprocessing, including data cleaning and data normalization. Data cleaning includes filtering low-quality genes / cells, removing batch effects, removing cell cycle influences, and handling null values. This step preserves high-quality data while removing lower-quality cells and data features, ensuring the reliability of subsequent analyses. Data normalization supports two methods: the `Normalize` function and the `sctransform` function from the `seurat` package. Since technical differences may exist between batches of scRNA-seq data, this can cause confounding of results. `sctransform` adjusts the expression matrix through non-linear transformation, reducing batch effects and improving data consistency and comparability.

[0074] The preprocessed data are subjected to routine functional analysis of single-cell transcriptome, single-cell immune repertoire, and single-cell multimodal data. The functional analysis includes dimensionality reduction clustering, marker gene analysis, copy number variation analysis, cell annotation, KEGG / GO enrichment analysis, and multi-omics aggregation.

[0075] The results of the functional analysis are visualized, including heatmaps, violin plots, bar charts, spatial visualizations, MA plots, and scatter plots. Heatmaps are mainly used to show the expression of specific gene sets, violin plots and bar charts are used to show the statistical situation of specific genes in different cell groups, and MA plots and scatter plots are used to show the results of differential gene screening in different cell groups.

[0076] Example 3

[0077] A functional analysis module based on multi-platform single-cell data, using the Singularity container engine.

[0078] The process is as follows:

[0079] Install the Singularity container engine on the host where the image was created;

[0080] Write the Singularity image definition file. This is a text file that predefines the base image used for building the image, the commands executed during the build process, etc. The detailed steps for writing the file are as follows:

[0081] (1) Load the CentOS 7.0 base image based on the Docker image source using Bootstrap;

[0082] (2) Install the dependency library files related to CentOS 7.0 system;

[0083] (3) Install the MambaForge software, and then use the Mamba software to quickly install Python, R base software and various single-cell data analysis software (such as Seurat, Scanpy, Monocle, Infercnv, etc.);

[0084] (4) Install the modular toolkit OESingleCell for single-cell multi-omics analysis from a private GitLab remote repository using pip+install;

[0085] (5) Install Jupyterlab and related plugins via conda, and configure the kernel and security verification information. The kernel is compatible with multiple versions of R kernel and Python kernel, and can be flexibly configured based on actual production needs.

[0086] (6) Set the default environment of the image through the environment parameter, and create the data path and working path to be mounted inside the container.

[0087] The Singularity container engine uses the `build` command to read the Singularity image definition file and build the binary image file.

[0088] On the cluster where this production environment needs to be deployed, copy the binary image file created above. When the image is loaded by executing the singularity's exec or shell command with a non-root account, the generated container will directly inherit the system user permissions and network settings, without needing to be set separately. This operation is more convenient and avoids the data security risks that may be caused by using root privileges.

[0089] Example 4

[0090] A modular tool for single-cell multi-omics analysis based on the Snakemake workflow engine, combined with S1.

[0091] With the support of the Singularity container engine, single-cell transcriptome project analysis reports are generated, forming a complete single-cell omics data analysis workflow. The steps are as follows:

[0092] Install the snakemake workflow engine on the host machine where task scheduling is performed;

[0093] Based on the modular tool for single-cell transcriptome analysis, using the Sankemake syntax, Snakemake automated workflow files are written to define the standard single-cell multi-omics analysis tasks to be performed. The main files involved include: project configuration files and analysis module rule files.

[0094] Project configuration file: Covers all parameters used in the execution of the Snakemake automated process, including basic project information (such as project number, species, original data address, etc.), control over whether the analysis is performed (such as whether to perform quality control, dimensionality reduction clustering, report generation, etc.), key parameters corresponding to the modular tools for single-cell multi-omics analysis (such as the selection of specific methods for modules such as batch processing and dimensionality reduction clustering), and selection of reference databases (such as reference genomes, etc.).

[0095] Analysis module rule file: This part is the main part of the Snakemake automation process. It defines the various sub-tasks of the single-cell multi-omics standard analysis task through some key parameters. The input parameter is used to configure the input file, the output parameter is used to configure the output file, the singularity parameter is used to configure the image file path, the resources parameter is used to specify the running resources (including memory and CPU), and the benchmark parameter is used to specify the task monitoring log output.

[0096] During the analysis, the system can pull the previously built and complete Snakemake automated workflow files from the private GitLab remote repository. After completing the environment initialization configuration and basic project information configuration, and submitting the task execution command, the system will automatically submit the task to the high-performance computing cluster node to complete the single-cell multi-omics standard analysis tasks defined in advance by the project configuration file, such as alignment, quality control, dimensionality reduction clustering, feature gene screening, marker gene identification, etc., and finally obtain a single-cell multi-omics standard analysis report.

[0097] Example 5

[0098] An interactive analysis platform for JupyterLab built using Singularity containerization technology.

[0099] The process of conducting exploratory analysis of single-cell data by multiple users using the JupyterLab interactive analysis platform involves the following steps:

[0100] When each sub-functional program module finishes running successfully, the source code is automatically converted into a JupyterLab executable file notebook through the Jupyter extension plugin jupytext, maintaining consistency between the exploratory data analysis methods and the automated analysis workflow code;

[0101] When users have exploratory analysis needs, they can start the JupyterLab web server through singularity to perform exploratory analysis. The specific startup command is as follows: `singularity exec --bind / path / outside / data: / path / inside / data --bind / path / outside / notebook: / path / inside / notebook --bind / path / outside / workspace: / path / outside / workspace build.sif jupyter lab --notebook-dir= / path / to / dir`` --no-browser --port=${port} --ip=${node}`

[0102] This means starting the JupyterLab web server using the exec command based on Singularity. During the startup process, the working directory is mounted to the specified location in the container using the bind parameter, and the IP address and port of the JupyterLab web server are specified. The --no-browser parameter is also added to prevent the browser from loading automatically.

[0103] Among them: the bind parameter can mount user data from outside the container into the container; the port parameter can set the listening port of the JupyterLab server; the ip parameter can specify the listening address of the JupyterLab server, which is usually the host machine's IP address; the no-browser parameter controls whether a browser is opened.

[0104] Enter the JupyterLab server address in your browser to access the JupyterLab server login interface. Enter your password and pass the security verification to enter the JupyterLab server. Through the use of mirrors and port allocation, each user's JupyterLab service can be independent and will not interfere with each other.

[0105] JupyterLab Server, as a web-based interactive analysis platform, supports various functions such as notebook operation, terminal operation, console interaction, text editing, and image viewing, providing a foundation for various operations in exploratory analysis of single-cell data;

[0106] By adopting the Singularity container engine, the process of automating the deployment of a single-cell multi-omics data analysis environment is configured with components that can execute various user codes. It supports mainstream programming languages ​​such as R and Python in single-cell data analysis, enabling rapid switching between different language analysis environments and demonstrating outstanding application effects in single-cell data analysis.

[0107] Advantages of this invention: Through the above technical solution, this invention...

[0108] (1) Based on the modular tool for single-cell transcriptome analysis, the problem of source data processing for various single-cell technology platforms is solved. By generating a unified data storage medium, the data acquisition stage in the analysis process is accelerated, and various operations on the data storage medium are supported. The modular toolkit has a unified interface, and the functions of each analysis module can be implemented independently through sub-commands, which facilitates the further development and expansion of the analysis tool. At the same time, the toolkit realizes the real-time saving of analysis scripts, running parameters, environment configuration parameters, running logs, etc., ensuring the traceability of the single-cell transcriptome data analysis process and the reproducibility of data analysis results.

[0109] (2) Based on Singularity container technology, it solves the problems of large deployment and maintenance difficulty and complexity caused by many dependent software and chaotic version numbers in traditional technical solutions: On the one hand, Singularity is compatible with Docker images, which is convenient to build, and the built images can be directly copied and ported to HPC clusters with Singularity installed for reuse, which saves environment configuration costs and ensures the reproducibility of data analysis results; On the other hand, Singularity supports running on systems without root privileges, which takes into account data security and convenience, and is more in line with the real scenario of single-cell bioinformatics analysis.

[0110] (3) Based on the Snakemake workflow engine technology, modular tools are combined to form an automated workflow for single-cell transcriptome data. It can be used with high-performance computing cluster task scheduling system to realize the automated generation of basic data analysis reports. It also supports Singularity environment image, which has a high advantage in workflow upgrade, optimization and migration.

[0111] (4) By enabling the JupyterLab web service system, the actual needs of multiple users on the high-performance computing cluster to conduct scientific exploration and analysis of single-cell data were met.

[0112] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for constructing a single-cell multi-omics data analysis system using containerization technology, characterized in that, The method for constructing the single-cell multi-omics data analysis system includes the following steps: Develop a modular toolkit for single-cell multi-omics analysis; The production environment is deployed by building production images on a high-performance cluster using the Singularity container engine. A single-cell multi-omics automated analysis platform was built using the snakemake workflow engine. Develop an interactive JupyterLab analysis platform for exploratory analysis of single-cell multi-omics data; Modular toolkits, automated analysis platforms, and interactive analysis platforms are used to build a single-cell multi-omics data analysis system through containerization and process management technologies.

2. The method for constructing a single-cell multi-omics data analysis system according to claim 1, characterized in that, The modular toolkit for single-cell multi-omics analysis contains eight modular tools.

3. The method for constructing a single-cell multi-omics data analysis system according to claim 2, characterized in that, The eight modular tools include: a data preprocessing and quality control module; a data integration and batch correction module; a dimensionality reduction, clustering, and cell annotation module; a multi-omics data integration and joint analysis module; a trajectory inference and cell dynamics module; a gene regulatory network analysis module; a functional enrichment analysis module; and an intercellular communication analysis module.

4. The method for constructing a single-cell multi-omics data analysis system according to claim 2, characterized in that, The modular toolkit is compiled using the R language and can be invoked via command line in a Linux environment or by loading R language packages in the R kernel of the JupyterLab interactive analysis platform. The modular toolkit can quickly perform data storage medium creation, updating, extraction, downsampling, and conversion operations by calling subroutine modules.

5. The method for constructing a single-cell multi-omics data analysis system according to claim 4, characterized in that, The modular toolkit can create data storage media based on processing data from different single-cell platforms; the update refers to updating existing data storage media and adding new feature data; the extraction refers to extracting sub-data from existing data storage media according to analysis requirements, that is, extracting data related to analysis requirements from existing data; the downsampling refers to randomly sampling existing data storage media to reduce the amount of data without affecting the analysis results; and the transformation refers to converting the format of data storage media according to analysis requirements.

6. The method for constructing a single-cell multi-omics data analysis system according to claim 1, characterized in that, The modular toolkit for single-cell multi-omics analysis is built using a development directory, which contains multiple folders, including: R folder, exec folder, man folder, and data folder.

7. The method for constructing a single-cell multi-omics data analysis system according to claim 6, characterized in that, The R folder is used to store general basic and functional functions for single-cell multi-omics analysis; the exec folder is used to store executable files, including the main program module files sctool and scVis and their corresponding subroutine modules; the man folder is used to store R markdown documents explaining the use of functions in R; and the data folder is used to store the data sets involved in the processing of each module in the toolkit.

8. The method for constructing a single-cell multi-omics data analysis system according to claim 1, characterized in that, The process of building a production image on a high-performance cluster using the Singularity container engine to complete the deployment of the production environment includes the following steps: Install the Singularity container engine on the host where the image was created; Write a Singularity image definition file, which predefines the base image used to build the image and the commands executed during the build process; The Singularity container engine uses the `build` command to read the Singularity image definition file and build the binary image file. On the cluster where the production environment needs to be deployed, copy the binary image file created above to complete the deployment of the production environment.

9. The method for constructing a single-cell multi-omics data analysis system according to claim 1, characterized in that, The construction of a single-cell multi-omics automated analysis platform using the Snakemake workflow engine includes the following steps: Install the snakemake workflow engine; Based on the pre-built modular toolkit for single-cell multi-omics analysis, the Snakemake automation process file is written using the Sankemake syntax to define the standard single-cell multi-omics analysis tasks to be performed. Pull the complete Snakemake automation workflow files, and after completing the environment initialization configuration and project basic information configuration, obtain the single-cell multi-omics automated analysis platform.

10. The method for constructing a single-cell multi-omics data analysis system according to claim 1, characterized in that, The JupyterLab interactive analysis platform for exploratory analysis of single-cell multi-omics data includes automatically converting source code into JupyterLab executable notebook files via the jupytext plugin, maintaining consistency between exploratory data analysis methods and automated analysis workflow code.

Images