Single-cell tumor microenvironment data analysis method based on gene regulatory networks

Abstract

This invention discloses a method for analyzing single-cell tumor microenvironment data based on gene regulatory networks, comprising the following steps: acquiring raw data values ​​of several cells, performing preprocessing and identifying characteristic genes; constructing a degree gene characterization matrix based on the obtained regulatory relationships between characteristic genes, and identifying cell subpopulations; and performing cell entropy analysis, differential degree gene identification, and gene function enrichment analysis on the cell subpopulations. This invention uses the SCILE algorithm to assess the overall stemness entropy of the cell based on the importance of each gene in the gene regulatory network of each cell. Compared to traditional expression-based assessment methods, this method avoids the impact of highly expressed, low-regulated genes on cell stemness.

Description

Technical Field

[0001] This invention belongs to the field of single-cell genomics data analysis, and in particular relates to a method for analyzing single-cell tumor microenvironment data based on gene regulatory networks. Background Technology

[0002] In recent years, tumor immunotherapy has become a major research direction in cancer treatment. Among them, exploring the composition of tumor cells to reveal the changing patterns of the tumor microenvironment is a key strategy to support tumor immunotherapy research.

[0003] In recent years, single-cell sequencing technology has been widely applied in developmental biology, immunology, and cell biology. Since tumor development and progression are complex and dynamic homeostatic regulatory processes involving numerous metabolic reprogramming processes, accurately identifying tumor cells and their microenvironment cells remains a major challenge in single-cell tumor microenvironment research. While integrated tools such as Seruat and Scanpy have been widely used for identifying single-cell subsets, these tools currently focus on traditional expression levels as the primary method for cell identification, neglecting the essential differences in intrinsic regulatory mechanisms within cells. The emergence of gene regulatory networks has made it possible to identify cells from gene networks, avoiding data redundancy and loss of crucial data caused by expression levels, and enabling the integration of data from different batches. Therefore, proposing a single-cell tumor microenvironment data analysis method based on gene regulatory networks is of significant importance in single-cell tumor microenvironment research. Summary of the Invention

[0004] The purpose of this invention is to provide a method for analyzing single-cell tumor microenvironment data based on gene regulatory networks, so as to solve the problems existing in the prior art.

[0005] To achieve the above objectives, this invention provides a method for analyzing single-cell tumor microenvironment data based on gene regulatory networks, comprising the following steps:

[0006] The raw data values ​​of several cells were obtained and preprocessed, and characteristic genes were identified.

[0007] Based on the regulatory relationships between the obtained characteristic genes, a Degree gene characterization matrix was constructed, and cell subpopulations were identified.

[0008] Cell entropy analysis, differentially expressed degree genes identification, and gene function enrichment analysis were performed on the cell subpopulations.

[0009] Optionally, the original dataset consists of readcount values ​​for several cells, including cell information and transcriptional expression information.

[0010] Optionally, the process of identifying feature genes in the preprocessed raw data values ​​includes: identifying feature genes between cells based on the entropy algorithm, wherein the threshold of the feature genes satisfies Δentropy>2 and P<0.05; where P represents the significance of the entropy algorithm.

[0011] Optionally, the process of constructing the Degree gene characterization matrix includes: constructing a single-sample gene regulation network based on the CSN algorithm to obtain the regulatory relationship between each feature gene, then performing overall dimensionality reduction on the feature genes based on the PCA algorithm, and characterizing each feature gene using the Degree algorithm to obtain the Degree gene characterization matrix.

[0012] Optionally, the process of identifying cell subpopulations includes: using the Degree gene characterization matrix as background data, and performing unsupervised clustering based on the SNN algorithm to identify key cell subpopulations.

[0013] Optionally, the process of performing cell entropy analysis on the cell subpopulation includes: using the Degree gene characterization matrix as background data, and obtaining cell entropy values ​​for different cell subpopulations based on the SCILE algorithm.

[0014] Optionally, the process of identifying differentially related genes in the cell subpopulation includes: obtaining the log2(FC) and p-value of the feature gene based on the Wilcox test algorithm, and identifying differentially related genes with a preset threshold of |log2(FC)|>0.25 and p<0.05; wherein the p-value is the Wilcox test verification value.

[0015] Optionally, the process of performing gene function enrichment analysis includes: performing GO enrichment analysis on differentially expressed Degree genes in each cell subpopulation based on enrichGO.

[0016] The technical effects of this invention are as follows:

[0017] Current methods for identifying single cells primarily rely on gene expression distribution to determine the commonalities and specificities between cells. However, they fail to consider the commonalities and specificities of the actual internal operational logic within cells (i.e., single-cell gene regulatory relationships), thus hindering the objective description of cell subpopulation distribution. To address these issues, this invention acquires raw data from several cells, constructs a single-sample gene regulatory network using the CSN algorithm, and then uses the Degree algorithm to characterize the importance of each gene within each cell's gene regulatory network, constructing a cell characterization matrix. Subsequently, the SNN algorithm is used to identify cell subpopulations, and Wilcox identifies specific driver-differential genes within the cell population. GO enrichment analysis reveals the driving biological functions. Compared to traditional cell subpopulation identification methods based on expression levels, this invention constructs each sample as a complex regulatory network, identifying subpopulations based on the similarity of regulatory networks between different samples. This avoids grouping bias caused by uneven expression distribution. Furthermore, for differential gene identification, this method uses the Degree gene characterization matrix as background data, enabling the capture of genes with small expression differences but significant differences in regulatory patterns between groups—essentially, "dark matter" genes in the cell.

[0018] Tumor cells have a lower degree of differentiation than normal cells. To assess the stemness of tumor cells and their microenvironment, this invention uses the SCILE algorithm to evaluate the overall stemness entropy of cells based on the importance of each gene in the gene regulatory network of each cell. Compared with traditional expression-based assessment methods, this method avoids the impact of high-expression, low-regulated genes on cell stemness. Attached Figure Description

[0019] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings:

[0020] Figure 1 This is a flowchart of the single-cell tumor microenvironment data analysis method in an embodiment of the present invention;

[0021] Figure 2 This is a schematic diagram of the cell subpopulation identification results in an embodiment of the present invention;

[0022] Figure 3 This is a schematic diagram of the cell subpopulation entropy results in an embodiment of the present invention;

[0023] Figure 4 The results of GO enrichment analysis of cell subpopulations in this embodiment of the invention are shown. Detailed Implementation

[0024] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0025] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0026] Example 1

[0027] like Figure 1 As shown, this embodiment provides a method for analyzing single-cell tumor microenvironment data based on gene regulatory networks, including the following steps:

[0028] Step 1: Raw Data Preprocessing

[0029] The original dataset consists of read count values ​​for 33,129 genes and 9,000 cells obtained using 10X Genomics, and includes cell information and transcriptional expression information. This embodiment first creates a Seurat object from the count data, and then removes the influence of variance using the Scale function, implemented via scaleData().

[0030] Step 2: Characteristic Gene Identification

[0031] The intrinsic entropy algorithm was used to identify highly informative characteristic genes among cells, with a threshold of Δentropy > 2 and P < 0.05; P represents the significance of the entropy algorithm.

[0032] Step 3: Construct a Degree gene characterization matrix using a single-sample gene regulatory network and identify cell subsets.

[0033] Using the expression matrix of the characteristic gene in all cells as background data, a single-sample gene regulation network is constructed using the CSN algorithm, and then the overall gene dimension is reduced using the PCA algorithm. Then, the Degree gene representation matrix of the characteristic gene is calculated using the Degree algorithm. Finally, using the Degree gene representation matrix as background data, the SNN algorithm is used to perform unsupervised clustering to identify key cell subpopulations.

[0034] Step 4: Cell entropy analysis based on the Degree gene characterization matrix

[0035] Using the Degree gene characterization matrix of characteristic genes as background data, the entropy value of each cell is calculated using the SCILE algorithm to infer the degree of cell stemness.

[0036] Step 5: Differential Degree Gene Identification

[0037] Using the Degree gene characterization matrix of characteristic genes as background data, the Wilcox test algorithm was applied to select |log2(FC)|>0.25& and p<0.05 as the threshold (here, the p value is the validation p value of the Wilcox test analysis) to calculate the differential genes in each cell subpopulation.

[0038] Step 6: Gene Function Enrichment Analysis

[0039] For each module of differentially expressed Degree genes, enrichGO was used for GO enrichment analysis.

[0040] Example 2

[0041] Using the single-cell tumor microenvironment data analysis method based on gene regulatory networks described in Example 1, data analysis was performed on single-cell transcriptome data of multiple myeloma NK cells. The specific steps included:

[0042] (1) Preprocessing of raw data

[0043] Specifically, the following steps are included:

[0044] ① Download single-cell data of the multiple myeloma microenvironment from the GEO database (https: / / cancergenome.nih.gov / ). There are 6,145 NK cells in the multiple myeloma microenvironment, and each sample contains 33,660 genes.

[0045] ②Removing cells with a total gene expression count of less than 200 and genes from cells with a gene expression data of less than 5% resulted in 15,399 genes and 5,381 cells.

[0046] ③ Finally, the data is normalized using the scale function to obtain the preprocessed expression spectrum. Some of the preprocessed original data are shown in Table 1.

[0047] Table 1

[0048]

[0049] (2) Feature gene identification

[0050] ① The intrinsic entropy algorithm was used to calculate the Δentropy and P of all genes. Some results are shown in Table 2.

[0051] ② With a threshold of Δentropy>2 and P<0.05, a total of 3529 characteristic genes were identified.

[0052] Table 2

[0053] Gene entropy_tot P HMGB1 0.475796 0.049 PTPRC 0.459191 0.048 UBC 0.448198 0.047 NEAT1 0.441887 0.046 BTG1 0.424939 0.045 BTF3 0.411283 0.044 CALM1 0.381548 0.043 ARHGDIB 0.376661 0.042 COX4I1 0.379028 0.041 COX7C 0.380376 0.04 OAZ1 0.353457 0.039 CORO1A 0.343398 0.038 ARPC2 0.343904 0.037 JUNB 0.33509 0.036 UBB 0.316062 0.035 RPL17 0.32735 0.034 UQCRB 0.319811 0.033 GAPDH 0.294536 0.032 ID2 0.293882 0.031 CTSW 0.305125 0.03

[0054] (3) A Degree gene characterization matrix was constructed using a single-sample gene regulation network, and cell subpopulations were identified.

[0055] like Figure 2 As shown, the specific steps include:

[0056] ① For each characteristic gene, the CSN algorithm is used to calculate the regulatory relationship between genes, and each gene is characterized by Degree to obtain the Degree gene characterization matrix. Partial results of the Degree characterization matrix of nk cells in multiple myeloma are shown in Table 3.

[0057] ② Next, firstly, the PCA algorithm is applied to reduce the dimensionality of the feature genes, and then the SNN algorithm is used to identify key cell subpopulations, resulting in 13 NK cell subpopulations.

[0058] ③ Dimensionality reduction of cell PCA results was performed using UMAP, with a UMAP threshold of 1:10. To visually represent the cell subpopulation results, the "DimPlot" function was used to display the UMAP results. See [link to results]. Figure 2 .

[0059] Table 3

[0060] Cell 1 Cell 2 Cell 3 Cell 4 Cell 5 Cell 6 RP11-34P13.3 0 0 0 0 0 0 FAM138A 0 0 0 0 0 0 OR4F5 0 0 0 0 0 0 RP11-34P13.7 0 0 0 0 0 0 RP11-34P13.8 0 0 0 0 0 0 RP11-34P13.14 0 0 0 0 0 0 RP11-34P13.9 0 0 0 0 0 0 FO538757.3 0 0 0 0 0 0 FO538757.2 0 1 0 0 0 0 AP006222.2 0 0 0 0 0 0 RP5-857K21.15 0 0 0 1 0 0 RP4-669L17.2 0 0 0 0 0 0 RP4-669L17.10 0 0 0 0 0 0 OR4F29 0 0 0 0 0 0 RP5-857K21.4 0 3 0 0 0 3 RP5-857K21.2 0 0 0 4 0 0 OR4F16 0 0 0 0 0 0 RP11-206L10.4 0 2 0 2 0 0 RP11-206L10.9 0 0 0 0 0 0 FAM87B 0 0 0 0 0 0

[0061] (4) Cell entropy analysis based on the Degree gene characterization matrix

[0062] like Figure 3 Specifically, it includes the following steps:

[0063] ① The SCILE algorithm was used to calculate the entropy value of each cell for the characteristic gene corresponding to the Degree gene representation matrix. To visually demonstrate the distribution of cell entropy values, the "boxplot" function was used to plot the entropy distribution of different cell subpopulations. Some results are shown below. Figure 2 .

[0064] (5) Differential Degree Gene Identification

[0065] Specifically, the following steps are included:

[0066] ① The log2(FC) and P-value of all feature genes were calculated using the Wilcox algorithm. Some results are shown in Table 4.

[0067] ② Set the constraint |log2(FC)|>0.25, P<0.05 to identify differentially expressed degree genes. Cluster 0 obtained 146 differentially expressed genes; cluster 1 obtained 202 differentially expressed genes; cluster 2 obtained 305 differentially expressed genes; cluster 3 obtained 197 differentially expressed genes; cluster 4 obtained 209 differentially expressed genes; cluster 5 obtained 316 differentially expressed genes; cluster 6 obtained 126 differentially expressed genes; cluster 7 obtained 262 differentially expressed genes; cluster 8 obtained 321 differentially expressed genes; cluster 9 obtained 282 differentially expressed genes; cluster 10 obtained 277 differentially expressed genes; cluster 11 obtained 516 differentially expressed genes; and cluster 12 obtained 216 differentially expressed genes. Table 4 shows the results of the differentially expressed degree gene analysis.

[0068] Table 4

[0069]

[0070]

[0071] (6) Gene function enrichment analysis

[0072] like Figure 4 Specifically, it includes the following steps:

[0073] ① Enrichment analysis was performed on the differentially expressed Degree genes of each cell subpopulation with biological processes (BP) provided by the GO database, and the 10 terms with the smallest p-values ​​were selected for further study. Results for some cell subpopulations are shown below. Figure 4 As shown.

[0074] Gene function enrichment analysis revealed that the differential degree of cell subset cluster0 is mainly involved in viral regulation, non-specific immune response, and tumor necrosis factor-related signaling pathway regulation, and is closely related to the NK cell immune regulation pattern in myeloma. This further corroborates the significant role of single-cell tumor microenvironment data analysis method based on single-sample gene regulatory network in identifying cell subsets in the tumor microenvironment.

[0075] The tumor immune microenvironment is a key challenge in single-cell research, particularly in identifying single-cell subpopulations. Current methods primarily rely on gene expression distribution to determine commonalities and specificities among cells, neglecting the commonalities and specificities of the actual internal operational logic within the cell (i.e., single-cell gene regulatory relationships) to objectively describe cell subpopulation distribution. To address this gap, this embodiment uses 10X single-cell expression data as background data. A single-sample gene regulatory network is constructed using the CSN algorithm. Then, the Degree algorithm is used to characterize the importance of each gene within the gene regulatory network of each cell, constructing a cell characterization matrix. Subsequently, the SNN algorithm is used to identify cell subpopulations, and Wilcox algorithm is used to identify specific driver genes within the cell population. GO enrichment analysis is then employed to reveal their driving biological functions.

[0076] Tumor cells have a lower degree of differentiation than normal cells. To assess the stemness of tumor cells and their microenvironment, this embodiment uses the SCILE algorithm to evaluate the overall stemness entropy of the cell based on the importance of each gene in the gene regulatory network of each cell. Compared with traditional evaluation methods based on expression levels, the method in this embodiment avoids the impact of high expression and low regulation of genes on cell stemness.

[0077] The above description is merely a preferred embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for analyzing single-cell tumor microenvironment data based on gene regulatory networks, characterized in that, Includes the following steps: The raw data values ​​of several cells were obtained and preprocessed, and characteristic genes were identified. Based on the regulatory relationships between the obtained characteristic genes, a Degree gene characterization matrix was constructed, and cell subpopulations were identified. Cell entropy analysis, differentially expressed degree genes identification, and gene function enrichment analysis were performed on the cell subpopulations. The process of constructing the Degree gene characterization matrix includes: constructing a single-sample gene regulation network based on the CSN algorithm to obtain the regulatory relationship between each feature gene, then performing overall dimensionality reduction on the feature genes based on the PCA algorithm, and characterizing each feature gene through the Degree algorithm to obtain the Degree gene characterization matrix. The process of identifying cell subpopulations includes: using the Degree gene characterization matrix as background data, and performing unsupervised clustering based on the SNN algorithm to identify key cell subpopulations; The process of performing cell entropy analysis on the cell subpopulations includes: using the Degree gene characterization matrix as background data, and obtaining cell entropy values ​​for different cell subpopulations based on the SCILE algorithm; The process of identifying differentially related genes in the cell subpopulation includes: obtaining the log2(FC) and p-value of the feature gene based on the Wilcox test algorithm, and identifying differentially related genes with a preset threshold of |log2(FC)|>0.25 and p<0.05; wherein, the p-value is the Wilcox test analysis verification value; The process of gene function enrichment analysis includes: performing GO enrichment analysis on differentially expressed Degree genes in each cell subpopulation based on enrichGO.

2. The method for analyzing single-cell tumor microenvironment data based on gene regulatory networks according to claim 1, characterized in that, The original dataset consists of readcount values ​​for several cells, including cell information and transcriptional expression information.

3. The method for analyzing single-cell tumor microenvironment data based on gene regulatory networks according to claim 1, characterized in that, The process of identifying characteristic genes in the preprocessed raw data values ​​includes: identifying characteristic genes between cells based on the entropy algorithm, wherein the threshold of the characteristic genes satisfies Δentropy>2 and p<0.05; where p represents the significance of the entropy algorithm.

Images