Cell subset division optimization method and device based on single cell clustering result
By performing initial clustering and differential expression analysis on single-cell RNA sequencing data, and iteratively merging cell subpopulations below a threshold, the subjectivity and inefficiency caused by excessive clustering of cell subpopulations in traditional methods are solved, achieving more accurate and efficient cell subpopulation segmentation.
Patent Information
- Application Number
- CN202511715083.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-11-21
- Publication Date
- 2026-03-03
AI Technical Summary
In traditional single-cell RNA sequencing data analysis, excessive clustering of cell subpopulations leads to redundant subpopulations that require manual merging. The lack of a unified data-driven standard results in highly subjective, poorly reproducible, and inefficient analysis results.
By performing initial clustering on single-cell RNA sequencing data, differential expression analysis is conducted to determine the number of differentially expressed genes. Cell subpopulations below a threshold are iteratively merged until the number of differentially expressed genes in all subpopulations reaches a preset threshold, and the optimized cell subpopulation division results are output.
It improves the accuracy and efficiency of cell subpopulation classification, and enhances the biological rationality and interpretability of the analysis results.
Smart Images
Figure CN121601027A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of medical data analysis technology, and in particular to a method and apparatus for optimizing cell subpopulation division based on single-cell clustering results. Background Technology
[0002] Single-cell RNA sequencing is a key technology for understanding cellular heterogeneity, and one of the core steps in its data analysis is cell clustering, which aims to divide cells into different subpopulations based on gene expression characteristics. However, traditional analytical workflows reveal significant limitations when conducting in-depth subpopulation analysis of specific cell types. After selecting an initial clustering resolution, analysts often obtain too many preliminary subpopulations with blurred boundaries. Some of these subpopulations exhibit highly similar expression patterns and lack clear, distinguishable biological characteristics, potentially leading to over-clustering.
[0003] Currently, to address the problem of over-clustering of subpopulations, researchers typically rely on manually merging redundant subpopulations. This method heavily depends on the researcher's professional experience and visual observation of the dimensionality reduction graph, lacking a unified, data-driven objective standard. This results in highly subjective analysis results with poor reproducibility. Furthermore, manually merging subpopulations requires analysts to continuously modify cell labels, recalculate differentially expressed genes, and perform visual verification, significantly reducing analytical efficiency. Summary of the Invention
[0004] This application provides a method and apparatus for optimizing cell subpopulation division based on single-cell clustering results, in order to solve the problems of high subjectivity, poor repeatability and low efficiency caused by the need for manual merging of redundant subpopulations generated by over-clustering.
[0005] In a first aspect, this application provides a method for optimizing cell subpopulation division based on single-cell clustering results, including:
[0006] Cluster analysis was performed on single-cell RNA sequencing data based on the preset initial clustering resolution to obtain multiple cell subpopulations after preliminary clustering;
[0007] Differential expression analysis was performed on each cell subpopulation obtained from the initial clustering to obtain the number of differentially expressed genes corresponding to each cell subpopulation;
[0008] Determine whether there are any cell subpopulations with a number of differentially expressed genes below a preset threshold among all cell subpopulations obtained from the initial clustering;
[0009] If any of the cell subpopulations obtained from the initial clustering have a number of differentially expressed genes below a preset threshold, then the cell subpopulations obtained from the initial clustering are iteratively merged until the number of differentially expressed genes in all merged cell subpopulations is not below the preset threshold, and the optimized cell subpopulation division result is output.
[0010] Optionally, differential expression analysis is performed on each cell subpopulation obtained from the initial clustering to obtain the number of differentially expressed genes corresponding to each cell subpopulation, including:
[0011] For any given target cell subpopulation, obtain the expression distribution values of the gene to be tested in the target cell subpopulation and the expression distribution values of the gene to be tested in the other cell subpopulations.
[0012] The obtained expression value distribution was subjected to between-group statistical tests to obtain the original test values, and the original test values were corrected by multiple tests to obtain the adjusted test values.
[0013] Determine the average fold change in expression of the gene to be tested relative to the other cell subpopulations in the target cell subpopulation;
[0014] Determine whether the gene to be tested meets the preset conditions, identify the gene to be tested that meets the preset conditions as the differentially expressed gene of the target cell subpopulation, and output a complete list of differentially expressed genes.
[0015] The preset conditions include: the gene to be tested is an upregulated gene in the target cell subpopulation; the proportion of cells expressing the gene to be tested in the target cell subpopulation is higher than a preset expression proportion threshold; the average fold change of the gene to be tested in the target cell subpopulation relative to other cell subpopulations is higher than a preset fold change threshold; and the adjusted test value is lower than a preset significance threshold.
[0016] Optionally, if any of the cell subpopulations obtained from the initial clustering contain a subpopulation with a number of differentially expressed genes below a preset threshold, then the cell subpopulations obtained from the initial clustering are iteratively merged, including:
[0017] The average gene expression profile of each cell subpopulation was determined, and a correlation matrix was constructed based on the average gene expression profile of the cell subpopulation.
[0018] Hierarchical clustering methods were used to construct expression profile clustering tree structures among cell subpopulations based on the correlation matrix;
[0019] From the constructed expression profile clustering tree structure, identify and merge the shortest branches that contain only two leaf nodes, where the leaf nodes represent cell subpopulations.
[0020] Optionally, a correlation matrix can be constructed based on the average gene expression profiles of cell subpopulations, including:
[0021] The average gene expression profiles of all cell subpopulations were aggregated to obtain the average expression level of each cell subpopulation.
[0022] Based on the average expression level of each cell subpopulation, the Pearson correlation coefficient between cell subpopulations is determined, and a symmetrical correlation matrix is formed based on the Pearson correlation coefficient between cell subpopulations.
[0023] Optionally, the average gene expression profiles of all cell subpopulations are aggregated to obtain the average expression level of each cell subpopulation, satisfying the following formula:
[0024]
[0025] in, For genes In cell subsets The average expression level in Cell subsets The total number of cells contained therein Representative cell subsets One specific cell, For genes In cells The log-normalized representation of the value.
[0026] Optionally, the Pearson correlation coefficient between cell subpopulations is determined based on the average expression level of each cell subpopulation, satisfying the following formula:
[0027]
[0028] in, Cell subsets and cell subsets The Pearson correlation coefficient between them This represents the total number of genes included in the average expression level. For genes In cell subsets The average expression level in For genes In cell subsets The average expression level in It is a cell subset average expression level All The mean of the average expression levels of each gene. It is a cell subset average expression level All The mean of the average expression level of each gene.
[0029] Optionally, a hierarchical clustering method is used to construct an expression profile clustering tree structure among cell subpopulations based on the correlation matrix, including:
[0030] The distance between cell subpopulations is determined based on the Pearson correlation coefficient between cell subpopulations, and the correlation matrix is converted into a distance matrix based on the distance between cell subpopulations.
[0031] Each cell subpopulation is treated as an independent cluster. The two clusters with the smallest distance values are found in the distance matrix. The two closest clusters are merged into a new cluster. The two merged old clusters are removed from the distance matrix, and the new distance between the new cluster and the other remaining clusters is calculated. The merging process is repeated until all cell subpopulations are merged into a single root cluster. The expression profile clustering tree structure between cell subpopulations is formed according to the merging order and distance.
[0032] Secondly, this application provides a cell subpopulation segmentation optimization device based on single-cell clustering results, the device comprising:
[0033] The processing module is used to perform cluster analysis on single-cell RNA sequencing data according to a preset initial clustering resolution to obtain multiple cell subpopulations after preliminary clustering.
[0034] The analysis module is used to perform differential expression analysis on each cell subpopulation obtained from the initial clustering, and to obtain the number of differentially expressed genes corresponding to each cell subpopulation.
[0035] The determination module is used to determine whether there are cell subpopulations with a number of differentially expressed genes below a preset threshold among all cell subpopulations obtained from the initial clustering;
[0036] The processing module is further configured to iteratively merge the cell subpopulations obtained from the initial clustering if there are cell subpopulations with a number of differentially expressed genes lower than a preset threshold among all cell subpopulations obtained from the initial clustering, until the number of differentially expressed genes in all merged cell subpopulations is not lower than the preset threshold, and output the optimized cell subpopulation division result.
[0037] Thirdly, this application provides a cell subpopulation segmentation optimization device based on single-cell clustering results, comprising:
[0038] Memory;
[0039] processor;
[0040] The memory stores computer-executed instructions;
[0041] The processor executes computer execution instructions stored in the memory to implement the cell subpopulation partitioning optimization method based on single-cell clustering results as described in the first aspect and various possible implementations of the first aspect above.
[0042] Fourthly, this application provides a computer storage medium having a computer program stored thereon, the computer program being executed by a processor to implement the cell subpopulation partitioning optimization method based on single-cell clustering results as described in the first aspect and various possible implementations of the first aspect above.
[0043] This application provides a method and apparatus for optimizing cell subpopulation segmentation based on single-cell clustering results. The method involves performing cluster analysis on single-cell RNA sequencing data according to a preset initial clustering resolution to obtain multiple cell subpopulations after preliminary clustering; performing differential expression analysis on each cell subpopulation to obtain the number of differentially expressed genes corresponding to each subpopulation; determining whether any cell subpopulations obtained from the preliminary clustering have a number of differentially expressed genes below a preset threshold; if such subpopulations exist, the cell subpopulations are iteratively merged until the number of differentially expressed genes in all merged subpopulations is not lower than the preset threshold, and then outputting the optimized cell subpopulation segmentation result. This method improves the accuracy of cell subpopulation segmentation and the efficiency of single-cell data analysis, enhancing the biological rationality and interpretability of the analysis results. Attached Figure Description
[0044] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.
[0045] Figure 1 A flowchart illustrating the cell subpopulation partitioning optimization method based on single-cell clustering results provided in this application embodiment;
[0046] Figure 2 This application provides a cell atlas of single-cell data after initial clustering in an embodiment of the application.
[0047] Figure 3 This is a cell atlas of fibroblast subpopulations provided in the embodiments of this application before annotation;
[0048] Figure 4 A schematic diagram of the correlation matrix and hierarchical clustering tree between fibroblast subpopulations before and after iteration, provided in an embodiment of this application;
[0049] Figure 5 The cell atlas of fibroblast subpopulations provided in the embodiments of this application is an automatically merged cell map;
[0050] Figure 6 Heatmap of the top 20 differentially expressed genes and enrichment of biological pathways in the fibroblast subpopulations provided in the embodiments of this application;
[0051] Figure 7This is an annotated cell atlas of fibroblast subpopulations provided in the embodiments of this application;
[0052] Figure 8 Cellular atlases of original cell subpopulations provided in embodiments of this application;
[0053] Figure 9 A schematic diagram of the structure of the cell subpopulation partitioning optimization device based on single-cell clustering results provided in an embodiment of this application;
[0054] Figure 10 This is a schematic diagram of the structure of the cell subpopulation segmentation optimization device based on single-cell clustering results provided in an embodiment of this application.
[0055] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation
[0056] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.
[0057] The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein.
[0058] In this application, the terms "exemplary" or "for example" are used to indicate examples, illustrations, or descriptions. Any embodiment or design described as "exemplary" or "for example" in this application should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of terms such as "exemplary" or "for example" is intended to present the relevant concepts in a specific manner.
[0059] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will now be described with reference to the accompanying drawings.
[0060] Figure 1 This is a flowchart illustrating the cell subpopulation partitioning optimization method based on single-cell clustering results provided in an embodiment of this application. Figure 1 As shown, the cell subpopulation partitioning optimization method based on single-cell clustering results provided in this embodiment includes:
[0061] S1: Perform cluster analysis on single-cell RNA sequencing data according to the preset initial cluster resolution to obtain multiple cell subpopulations after preliminary clustering.
[0062] Specifically, the acquired single-cell RNA sequencing data is preprocessed; a cell proximity map is constructed based on the preprocessed single-cell RNA sequencing data; cells are clustered according to a preset resolution based on the cell proximity map, and cell types are annotated according to a set of marker genes to obtain multiple candidate cell populations; the target cell population is determined from the multiple candidate cell populations, and the target cell population is subjected to dimensionality reduction clustering to obtain multiple unannotated cell subpopulations.
[0063] The preprocessing includes operations such as quality control, standardization, feature selection, and dimensionality reduction. Quality control aims to filter low-quality cells based on parameters such as the total number of RNA molecules, the total number of genes, and the percentage of mitochondrial genes. Standardization aims to eliminate differences in sequencing depth, making the gene expression values of all cells comparable. Feature selection aims to screen for highly variable genes and reduce noise interference. Dimensionality reduction aims to project the data into a low-dimensional space using PCA or nonlinear methods.
[0064] Exemplarily, the data of 23 single-cell RNA sequencing samples in papillary thyroid carcinoma GSE184362 was obtained. Quality control was performed with 500 < total number of RNA molecules < 5000, 200 < total number of genes < 10000, and mitochondrial gene percentage < 10%, resulting in 151056 cells. Standardization was carried out using a data standardization function. 2000 highly variable genes were selected. The expression values of each highly variable gene were adjusted to a similar range through data scaling (for example, adjusted to a standard normal distribution with a mean of 0 and a variance of 1) to ensure that each gene had equal weight in the analysis. Dimensionality reduction was performed using the principal component analysis method, batch effects were removed using the Harmony method, and non-linear dimensionality reduction visualization was carried out using non-linear dimensionality reduction methods (such as UMAP / tSNE). A K-nearest neighbor graph was constructed using the dimensionality-reduced and corrected data. In this low-dimensional space, for each cell, the k cells closest to it were calculated and found, generating a K-nearest neighbor graph. In the K-nearest neighbor graph, each cell (node) was connected (by an edge) to its k nearest neighbors. Based on the results of the KNN graph, the connection strength between any two cells was recalculated to construct a cell neighbor graph. Clustering was performed based on the cell proximity graph with a resolution of 0.1, and the cells were annotated according to classical marker genes, and these cells were annotated as a total of 6 types of cells as shown in Figure 2 : Thyrocytes (thyroid cells), T&NK cells, B cells, Myeloid cells (myeloid cells), Endothelial cells (endothelial cells), Fibroblasts (fibroblasts). Fibroblasts were selected, and after reclustering and dimensionality reduction, 8 annotated cell subsets as shown in Figure 3 were obtained.
[0065] S2: Differential expression analysis was performed on each cell subset obtained from the preliminary clustering to obtain the number of differentially expressed genes corresponding to each cell subset.
[0066] Specifically, for any target cell subset, the expression distribution value of the待测 gene in the target cell subset and the expression distribution value of the待测 gene in the remaining cell subsets were obtained; the obtained expression value distributions were subjected to an inter-group difference statistical test to obtain the original test value, and the original test value was corrected for multiple testing to obtain the adjusted test value; the average expression fold change value of the待测 gene in the target cell subset relative to the remaining cell subsets was determined; it was judged whether the待测 gene met the preset conditions, and the待测 gene that met the preset conditions was used as the differentially expressed gene of the target cell subset, and a complete list of differentially expressed genes was output. The differentially expressed genes in the list of differentially expressed genes were counted to obtain the number of differentially expressed genes corresponding to each cell subset.
[0067] The preset conditions include: the gene to be tested is an upregulated gene in the target cell subpopulation; the proportion of cells expressing the gene to be tested in the target cell subpopulation is higher than the preset expression proportion threshold; the average fold change of the gene to be tested in the target cell subpopulation relative to the other cell subpopulations is higher than the preset fold change threshold; and the adjusted test value is lower than the preset significance threshold.
[0068] For example, the Wilcoxon rank-sum test is used to test the differences between groups to obtain the original test values. The original test values are then corrected by multiple tests to obtain the adjusted test values. This method is suitable for the non-normal distribution characteristics of single-cell expression data. A fold change threshold of 1 is set to ensure that the selected genes have significant expression differences. An expression ratio threshold of 25% is set to filter low-expression genes to ensure that the results are representative. Only genes upregulated in the target subgroup are retained to avoid ambiguity caused by downregulated genes. A significance threshold of 0.05 is set to ensure that the difference results are statistically significant.
[0069] S3: Determine whether there are any cell subpopulations with a number of differentially expressed genes below a preset threshold among all cell subpopulations obtained from the initial clustering.
[0070] For example, a threshold of 50 is set, meaning that the number of differentially expressed genes between each cell subpopulation must be greater than or equal to 50, indicating that there is heterogeneity between cell subpopulations. If the number of differentially expressed genes in a certain cell subpopulation is less than the set threshold, it means that the difference between a certain cell subpopulation and other cell subpopulations is not significant, and the relevant cell subpopulations need to be merged.
[0071] S4: If there are cell subpopulations with a number of differentially expressed genes below a preset threshold among all cell subpopulations obtained from the initial clustering, then iteratively merge the cell subpopulations obtained from the initial clustering until the number of differentially expressed genes in all merged cell subpopulations is not lower than the preset threshold, and output the optimized cell subpopulation division result.
[0072] Specifically, if any cell subpopulation obtained from the initial clustering has a number of differentially expressed genes below a preset threshold, then the cell subpopulations obtained from the initial clustering are iteratively merged, including:
[0073] S41: Determine the mean gene expression profile of each cell subpopulation and construct a correlation matrix based on the mean gene expression profiles of all cell subpopulations.
[0074] Specifically, the average gene expression profiles of all cell subpopulations are aggregated to obtain the average expression level of each cell subpopulation; based on the average expression level of each cell subpopulation, the Pearson correlation coefficient between cell subpopulations is determined, and a symmetrical correlation matrix is formed based on the Pearson correlation coefficient between cell subpopulations.
[0075] More specifically, the average gene expression profiles of all cell subpopulations are aggregated to obtain the average expression level of each cell subpopulation, which satisfies the following formula:
[0076]
[0077] in, For genes In cell subsets The average expression level in Cell subsets The total number of cells contained therein Representative cell subsets One specific cell, For genes In cells The log-normalized representation of the value.
[0078] The Pearson correlation coefficient between cell subpopulations is determined based on the average expression level of each cell subpopulation, satisfying the following formula:
[0079]
[0080] in, Cell subsets and cell subsets The Pearson correlation coefficient between them This represents the total number of genes included in the average expression level. For genes In cell subsets The average expression level in For genes In cell subsets The average expression level in It is a cell subset average expression level All The mean of the average expression levels of each gene. It is a cell subset average expression level All The mean of the average expression level of each gene.
[0081] S42: Using hierarchical clustering methods, construct expression profile clustering tree structures among cell subpopulations based on the correlation matrix.
[0082] Specifically, the distance between cell subpopulations is determined based on the Pearson correlation coefficient, and the correlation matrix is converted into a distance matrix based on the distance between cell subpopulations. Each cell subpopulation is treated as an independent cluster, and the two clusters with the smallest distance values are found in the distance matrix. The two closest clusters are merged into a new cluster, and the two merged old clusters are removed from the distance matrix. The new distance between the new cluster and the other remaining clusters is calculated. The merging process is repeated until all cell subpopulations are merged into a single root cluster. An expression spectrum clustering tree structure is formed based on the merging order and distance. The expression spectrum clustering tree structure can visualize the expression similarity relationship between cell subpopulations.
[0083] Understandably, the Pearson correlation coefficient is an indicator of similarity. Converting a correlation index to a distance index uses the following formula:
[0084]
[0085] in, The distance after conversion. This is the Pearson correlation coefficient.
[0086] S43: From the constructed expression spectrum clustering tree structure, identify and merge the shortest branch that contains only two leaf nodes.
[0087] Leaf nodes represent cell subpopulations.
[0088] Specifically, the `cutree` method is used to prune the tree structure layer by layer, searching for and identifying the smallest clustering subtree containing exactly two leaf nodes. The algorithm captures the first combination separated by pruning that contains only two cell subpopulations. This combination represents the two subpopulations with the closest clustering distance and the highest priority to be merged in the current level of clustering structure, i.e., the shortest branch.
[0089] For example, when a branch exists in the hierarchical clustering tree containing only two real cell subpopulations, the branch is automatically identified and the two subpopulations are merged to generate a new cell subpopulation label. Further, differential analysis is performed again on the cell subpopulations and the merged cell subpopulations until the number of differentially expressed genes in all cell subpopulations is not lower than a preset threshold. Figure 4 As shown, after 5 iterations, the initial 8 subpopulations were automatically merged into 3 stable subpopulations: C03, C00+C02+C01+C04, and C05+C06+C07, with DEGs of 207, 126, and 83 respectively, all exceeding the set threshold of 50. All subpopulations met the termination condition, and the iteration process ended. The distribution of the 3 optimized subpopulations on the tSNE map is shown below. Figure 5 As shown.
[0090] In an optional embodiment, after outputting the optimized cell subpopulation segmentation results, the method further includes: performing functional enrichment analysis on the optimized subpopulations, and biologically naming the subpopulations based on their differentially expressed genes or enriched functions.
[0091] For example, the specific naming process is as follows: (1) Preliminary naming is performed based on representative differentially expressed genes, and the first 20 differentially expressed genes in each group are extracted, such as Figure 6As shown, key marker genes were identified and named. C00+C02+C01+C04 was named ACTA2+CAF. This subset highly expresses ACTA2, a gene encoding α-smooth muscle actin, a classic marker of myofibroblasts. Myofibroblasts possess strong cell contractile capacity and the function of secreting extracellular matrix, supporting tumor growth and invasion during tumor progression by remodeling the matrix, increasing tissue stiffness, and promoting angiogenesis. Therefore, high expression of ACTA2 strongly suggests that this subset plays the role of myofibroblasts (myoCAF). C03 was named CFD+CAF. This subset highly expresses CFD, a gene encoding complement factor D, a key rate-limiting enzyme in the complement system's alternative pathway, closely related to inflammatory responses. In the tumor microenvironment, CFD secreted by fibroblasts can promote local inflammation, regulate immune cell function, and may affect tumor cell survival through paracrine pathways. Therefore, high expression of CFD indicates that this subset possesses pro-inflammatory properties and can be classified as inflammation-associated fibroblasts (iCAF). The C05+C06+C07 subset is named CD74+CAF. This subset highly expresses CD74, the invariant strand of MHC class II molecules, which plays a crucial role in antigen presentation, responsible for the correct folding of MHC class II molecules and guiding them into the antigen processing compartment. Traditionally, fibroblasts were thought to lack specialized antigen presentation functions, but increasing research shows that under specific microenvironmental stimuli, some tumor-associated fibroblasts can upregulate MHC class II molecules, thereby acquiring the ability to present antigens and directly communicate with T cells. Therefore, high expression of CD74 indicates that this subset possesses atypical immunomodulatory functions and can be defined as antigen-presenting fibroblasts (apCAF). (2) Based on the functional enrichment pathways, the top 20 differentially expressed genes from each subpopulation obtained from the differential analysis were extracted. Using the ClusterGVis package, the prepareDataFromscRNA function was called to convert the Seurat object and differential gene results into standardized input data required for enrichment analysis. The enrichCluster function was used to perform enrichment analysis on differentially expressed genes in different cell subpopulations. The human gene annotation database was selected, the analysis type was biological process, and the significance threshold of the enrichment pathway was set to 0.05. The top 5 most significant enriched pathways were extracted from each cell subpopulation. After enrichment, the visCluster function was used for display, such as... Figure 7As shown, the first five pathways in the CO0+C02+C01+C04 group are intercellular junction assembly, intercellular junction organization, epithelial cell differentiation involved in kidney development, vascular endothelial growth factor signaling pathway, and cell differentiation involved in kidney development. These pathways are highly consistent with the functions of myofibroblasts. "Intercellular junctions" are the basis for maintaining tissue structure and mechanical tension, reflecting the ability of myoCAF to reshape physical barriers; while the "vascular endothelial growth factor signaling pathway" directly points to its core functions of promoting tumor angiogenesis and delivering nutrients to tumors. The first five pathways in the CO3 group are elastic fiber components, extracellular matrix organization, extracellular structural organization, external capsule structure organization, and extracellular matrix assembly. These pathways are all related to the synthesis and remodeling of the extracellular matrix (ECM), which is a typical characteristic of iCAF in constructing a pro-inflammatory and pro-fibrotic microenvironment by secreting various cytokines and matrix proteins. The first five pathways of CO5+C06+C07 are, respectively, the processing and presentation of exogenous peptide antigens through MHC class II molecules, the processing and presentation of peptide antigens through MHC class II molecules, the processing and presentation of peptide or polysaccharide antigens through MHC class II molecules, the processing and presentation of exogenous peptide antigens, and the assembly of MHC class II molecule protein complexes. These pathways clearly indicate that the core biological characteristic of this subpopulation is MHC class II molecule-mediated antigen presentation, enabling it to play a unique immunomodulatory role in the tumor immune microenvironment. In summary, based on the function of key marker genes and their enrichment pathways, the three optimized subpopulations were finally named myofibroblasts (myoCAF), inflammation-associated fibroblasts (iCAF), and antigen-presenting fibroblasts (apCAF).
[0092] like Figure 7 As shown and Figure 8 As shown, compared with the analysis results of fibroblasts in the dataset in the original literature, the method of the present invention not only successfully identified and reproduced the two subpopulations iCAF and myoCAF reported therein, but also more accurately identified apCAF, a special fibroblast subpopulation with important immune functions in the tumor microenvironment.
[0093] This application provides a method for optimizing cell subpopulation segmentation based on single-cell clustering results. The method performs clustering analysis on single-cell RNA sequencing data according to a preset initial clustering resolution to obtain multiple cell subpopulations after preliminary clustering. Differential expression analysis is then performed on each cell subpopulation to obtain the number of differentially expressed genes corresponding to each subpopulation. It is determined whether any cell subpopulations obtained from the preliminary clustering have a number of differentially expressed genes below a preset threshold. If such subpopulations exist, the cell subpopulations are iteratively merged until the number of differentially expressed genes in all merged subpopulations is not lower than the preset threshold, and the optimized cell subpopulation segmentation result is output. This method improves the accuracy of cell subpopulation segmentation and the efficiency of single-cell data analysis, enhancing the biological rationality and interpretability of the analysis results.
[0094] Figure 9 This is a schematic diagram of the cell subpopulation segmentation optimization device based on single-cell clustering results provided in an embodiment of this application. Figure 9 As shown, the cell subpopulation segmentation optimization device 200 based on single-cell clustering results provided in this embodiment includes:
[0095] The processing module 201 is used to perform cluster analysis on single-cell RNA sequencing data according to a preset initial clustering resolution to obtain multiple cell subpopulations after preliminary clustering.
[0096] Analysis module 202 is used to perform differential expression analysis on each cell subpopulation obtained from the preliminary clustering to obtain the number of differentially expressed genes corresponding to each cell subpopulation;
[0097] The determination module 203 is used to determine whether there are cell subpopulations with a number of differentially expressed genes below a preset threshold among all cell subpopulations obtained from the preliminary clustering;
[0098] The processing module 201 is further configured to iteratively merge the cell subpopulations obtained from the initial clustering when there are cell subpopulations with a number of differentially expressed genes lower than a preset threshold among all cell subpopulations obtained from the initial clustering, until the number of differentially expressed genes in all merged cell subpopulations is not lower than the preset threshold, and output the optimized cell subpopulation division result.
[0099] The cell subpopulation partitioning optimization device based on single-cell clustering results provided in this embodiment can execute the cell subpopulation partitioning optimization method based on single-cell clustering results provided in the above method embodiment. Its implementation principle and technical effect are similar, and will not be described in detail here.
[0100] Figure 10This is a schematic diagram of the structure of a cell subpopulation segmentation optimization device based on single-cell clustering results provided in an embodiment of this application. Figure 10 As shown in the embodiment of this application, the cell subpopulation segmentation optimization device based on single-cell clustering results is provided. The cell subpopulation segmentation optimization device 300 based on single-cell clustering results includes: receiver 301, transmitter 302, processor 303 and memory 304.
[0101] Receiver 301 is used to receive instructions and data;
[0102] Transmitter 302 is used to send commands and data;
[0103] Memory 304 is used to store computer-executed instructions;
[0104] The processor 303 is used to execute computer execution instructions stored in the memory 304 to implement the various steps of the cell subpopulation partitioning optimization method based on single-cell clustering results in the above embodiments. For details, please refer to the relevant descriptions in the foregoing embodiments of the cell subpopulation partitioning optimization method based on single-cell clustering results.
[0105] Optionally, the memory 304 can be either standalone or integrated with the processor 303.
[0106] When the memory 304 is set up independently, the electronic device also includes a bus for connecting the memory 304 and the processor 303.
[0107] This application embodiment also provides a computer storage medium storing computer execution instructions. When the processor executes the computer execution instructions, it implements the cell subpopulation partitioning optimization method based on single-cell clustering results executed by the cell subpopulation partitioning optimization device based on single-cell clustering results described above.
[0108] It will be understood by those skilled in the art that all or some of the steps, systems, or apparatuses disclosed above, and their functional modules / units, can be implemented as software, firmware, hardware, or suitable combinations thereof. In hardware implementations, the division between functional modules / units mentioned in the above description does not necessarily correspond to the division of physical components; for example, a physical component may have multiple functions, or a function or step may be performed collaboratively by several physical components. Some or all physical components may be implemented as software executed by a processor, such as a central processing unit, digital signal processor, or microprocessor, or as hardware, or as an integrated circuit, such as an application-specific integrated circuit (ASIC). Such software may be distributed on a computer-readable medium, which may include computer storage media (or non-transitory media) and communication media (or transient media). As is known to those skilled in the art, the term computer storage media includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, program modules, or other data). Computer storage media include, but are not limited to, RAM, ROM, EEPROM, flash memory or other memory technologies, CD-ROM, digital versatile disc (DVD) or other optical disc storage, magnetic cartridges, magnetic tape, disk storage or other magnetic storage devices, or any other medium that can be used to store desired information and can be accessed by a computer. Furthermore, it is well known to those skilled in the art that communication media typically contain computer-readable instructions, data structures, program modules, or other data in modulated data signals such as carrier waves or other transmission mechanisms, and may include any information delivery medium.
[0109] Other embodiments of this application will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of this application that follow the general principles of this application and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of this application are indicated by the following claims.
[0110] It should be understood that this application is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this application is limited only by the appended claims.
Claims
1. A method for optimizing cell subpopulation division based on single-cell clustering results, characterized in that, The method includes: Cluster analysis was performed on single-cell RNA sequencing data based on the preset initial clustering resolution to obtain multiple cell subpopulations after preliminary clustering; Differential expression analysis was performed on each cell subpopulation obtained from the initial clustering to obtain the number of differentially expressed genes corresponding to each cell subpopulation; Determine whether there are any cell subpopulations with a number of differentially expressed genes below a preset threshold among all cell subpopulations obtained from the initial clustering; If any of the cell subpopulations obtained from the initial clustering have a number of differentially expressed genes below a preset threshold, then the cell subpopulations obtained from the initial clustering are iteratively merged until the number of differentially expressed genes in all merged cell subpopulations is not below the preset threshold, and the optimized cell subpopulation division result is output.
2. The method according to claim 1, characterized in that, Differential expression analysis was performed on each cell subpopulation obtained from the initial clustering to obtain the number of differentially expressed genes corresponding to each cell subpopulation, including: For any given target cell subpopulation, obtain the expression distribution values of the gene to be tested in the target cell subpopulation and the expression distribution values of the gene to be tested in the other cell subpopulations. The obtained expression value distribution was subjected to between-group statistical tests to obtain the original test values, and the original test values were corrected by multiple tests to obtain the adjusted test values. Determine the average fold change in expression of the gene to be tested relative to the other cell subpopulations in the target cell subpopulation; Determine whether the gene to be tested meets the preset conditions, identify the gene to be tested that meets the preset conditions as the differentially expressed gene of the target cell subpopulation, and output a complete list of differentially expressed genes. The preset conditions include: the gene to be tested is an upregulated gene in the target cell subpopulation; the proportion of cells expressing the gene to be tested in the target cell subpopulation is higher than a preset expression proportion threshold; the average fold change of the gene to be tested in the target cell subpopulation relative to other cell subpopulations is higher than a preset fold change threshold; and the adjusted test value is lower than a preset significance threshold.
3. The method according to claim 1, characterized in that, If any of the cell subpopulations obtained from the initial clustering contain a subpopulation with a number of differentially expressed genes below a preset threshold, then the cell subpopulations obtained from the initial clustering are iteratively merged, including: The average gene expression profile of each cell subpopulation was determined, and a correlation matrix was constructed based on the average gene expression profile of the cell subpopulation. Hierarchical clustering methods were used to construct expression profile clustering tree structures among cell subpopulations based on the correlation matrix; From the constructed expression profile clustering tree structure, identify and merge the shortest branches that contain only two leaf nodes, where the leaf nodes represent cell subpopulations.
4. The method according to claim 3, characterized in that, A correlation matrix was constructed based on the average gene expression profiles of cell subpopulations, including: The average gene expression profiles of all cell subpopulations were aggregated to obtain the average expression level of each cell subpopulation. Based on the average expression level of each cell subpopulation, the Pearson correlation coefficient between cell subpopulations is determined, and a symmetrical correlation matrix is formed based on the Pearson correlation coefficient between cell subpopulations.
5. The method according to claim 4, characterized in that, The average gene expression profiles of all cell subpopulations are aggregated to obtain the average expression level of each cell subpopulation, which satisfies the following formula: ； in, For genes In cell subsets The average expression level in Cell subsets The total number of cells contained therein Representative cell subsets One specific cell, For genes In cells The log-normalized representation of the value.
6. The method according to claim 4, characterized in that, The Pearson correlation coefficient between cell subpopulations is determined based on the average expression level of each cell subpopulation, satisfying the following formula: ； in, Cell subsets and cell subsets The Pearson correlation coefficient between them This represents the total number of genes included in the average expression level. For genes In cell subsets The average expression level in For genes In cell subsets The average expression level in It is a cell subset average expression level All The mean of the average expression levels of each gene. It is a cell subset average expression level All The mean of the average expression level of each gene.
7. The method according to claim 3, characterized in that, Using hierarchical clustering methods, an expression profile clustering tree structure is constructed among cell subpopulations based on the correlation matrix, including: The distance between cell subpopulations is determined based on the Pearson correlation coefficient between cell subpopulations, and the correlation matrix is converted into a distance matrix based on the distance between cell subpopulations. Each cell subpopulation is treated as an independent cluster. The two clusters with the smallest distance values are found in the distance matrix. The two closest clusters are merged into a new cluster. The two merged old clusters are removed from the distance matrix, and the new distance between the new cluster and the other remaining clusters is calculated. The merging process is repeated until all cell subpopulations are merged into a single root cluster. The expression profile clustering tree structure between cell subpopulations is formed according to the merging order and distance.
8. A cell subpopulation partitioning optimization device based on single-cell clustering results, characterized in that, The device includes: The processing module is used to perform cluster analysis on single-cell RNA sequencing data according to a preset initial clustering resolution to obtain multiple cell subpopulations after preliminary clustering. The analysis module is used to perform differential expression analysis on each cell subpopulation obtained from the initial clustering, and to obtain the number of differentially expressed genes corresponding to each cell subpopulation. The determination module is used to determine whether there are cell subpopulations with a number of differentially expressed genes below a preset threshold among all cell subpopulations obtained from the initial clustering; The processing module is further configured to iteratively merge the cell subpopulations obtained from the initial clustering if there are cell subpopulations with a number of differentially expressed genes lower than a preset threshold among all cell subpopulations obtained from the initial clustering, until the number of differentially expressed genes in all merged cell subpopulations is not lower than the preset threshold, and output the optimized cell subpopulation division result.
9. A cell subpopulation segmentation optimization device based on single-cell clustering results, characterized in that, include: Memory; processor; The memory stores computer-executed instructions; The processor executes computer execution instructions stored in the memory to implement the cell subpopulation partitioning optimization method based on single-cell clustering results as described in any one of claims 1-7.
10. A computer storage medium, characterized in that, The computer storage medium stores computer execution instructions, which, when executed by a processor, are used to implement the cell subpopulation partitioning optimization method based on single-cell clustering results as described in any one of claims 1-7.